Since many of you wrote in asking me to expand on some of the bullet points in the 'IIT-B 2012 Pre-Placements Career Guidance' presentation, here is a limited and rough transcript of the talk I gave last night at IIT-Bombay. Q&A not included
Stochastic Control/Reinforcement Learning for Optimal Market MakingAshwin Rao
Optimal Market Making is the problem of dynamically adjusting bid and ask prices/sizes on the Limit Order Book so as to maximize Expected Utility of Gains. This is a stochastic control problem that can be tackled with classical Dynamic Programming techniques or with Reinforcement Learning (using a market-learnt simulator)
Adaptive Multistage Sampling Algorithm: The Origins of Monte Carlo Tree SearchAshwin Rao
The document summarizes the Adaptive Multistage Sampling (AMS) algorithm, which is considered the "spiritual origin" of Monte Carlo Tree Search (MCTS). AMS is a generic simulation-based algorithm for solving finite-horizon Markov Decision Processes when the state space is very large. It works by allocating a fixed number of action selections per state over time steps, and using an explore-exploit formula to select actions during simulations. The estimates produced by AMS are asymptotically unbiased and converge to the optimal values.
This document provides an overview of Ashwin Rao's presentation on the fundamental theorems of asset pricing. It begins with an outline of the topics to be covered, including an intuitive understanding using a simple single-period setting. It then covers portfolios and arbitrage, defines the risk-neutral probability measure, and states the first fundamental theorem of asset pricing relating arbitrage and the existence of a risk-neutral measure. Derivatives, replicating portfolios, and hedges are introduced. The second fundamental theorem is presented relating market completeness to a unique risk-neutral measure. Derivatives pricing is discussed based on replication and risk-neutral measures. Examples of incomplete markets with multiple risk-neutral measures are briefly mentioned.
Evolutionary Strategies as an alternative to Reinforcement LearningAshwin Rao
This document introduces Evolutionary Strategies (ES), a black-box optimization technique inspired by natural evolution. ES frames optimization as maximizing the expected value of an objective function over a parameterized probability distribution of candidate solutions. It uses stochastic gradient ascent to update the distribution's parameters based on sampled solutions and their objective values. The document applies ES to solving Markov Decision Processes by sampling candidate policies and estimating the gradient of expected return. While simpler than reinforcement learning, ES can be effective for optimization problems and more robust than advanced policy gradient methods.
Principles of Mathematical Economics applied to a Physical-Stores Retail Busi...Ashwin Rao
Ashwin Rao presents mathematical models for inventory control and other optimization problems in retail. He describes classical inventory models like economic order quantity and the newsvendor problem. These problems involve stochastic optimization and can be modeled as Markov decision processes. Rao also discusses multi-item and multi-location inventory control problems faced in real-world retail and how they require solving large MDPs using techniques like dynamic programming and reinforcement learning.
Understanding Dynamic Programming through Bellman OperatorsAshwin Rao
Policy Iteration and Value Iteration algorithms are best understood by viewing them from the lens of Bellman Policy Operator and Bellman Optimality Operator
Stochastic Control/Reinforcement Learning for Optimal Market MakingAshwin Rao
Optimal Market Making is the problem of dynamically adjusting bid and ask prices/sizes on the Limit Order Book so as to maximize Expected Utility of Gains. This is a stochastic control problem that can be tackled with classical Dynamic Programming techniques or with Reinforcement Learning (using a market-learnt simulator)
Adaptive Multistage Sampling Algorithm: The Origins of Monte Carlo Tree SearchAshwin Rao
The document summarizes the Adaptive Multistage Sampling (AMS) algorithm, which is considered the "spiritual origin" of Monte Carlo Tree Search (MCTS). AMS is a generic simulation-based algorithm for solving finite-horizon Markov Decision Processes when the state space is very large. It works by allocating a fixed number of action selections per state over time steps, and using an explore-exploit formula to select actions during simulations. The estimates produced by AMS are asymptotically unbiased and converge to the optimal values.
This document provides an overview of Ashwin Rao's presentation on the fundamental theorems of asset pricing. It begins with an outline of the topics to be covered, including an intuitive understanding using a simple single-period setting. It then covers portfolios and arbitrage, defines the risk-neutral probability measure, and states the first fundamental theorem of asset pricing relating arbitrage and the existence of a risk-neutral measure. Derivatives, replicating portfolios, and hedges are introduced. The second fundamental theorem is presented relating market completeness to a unique risk-neutral measure. Derivatives pricing is discussed based on replication and risk-neutral measures. Examples of incomplete markets with multiple risk-neutral measures are briefly mentioned.
Evolutionary Strategies as an alternative to Reinforcement LearningAshwin Rao
This document introduces Evolutionary Strategies (ES), a black-box optimization technique inspired by natural evolution. ES frames optimization as maximizing the expected value of an objective function over a parameterized probability distribution of candidate solutions. It uses stochastic gradient ascent to update the distribution's parameters based on sampled solutions and their objective values. The document applies ES to solving Markov Decision Processes by sampling candidate policies and estimating the gradient of expected return. While simpler than reinforcement learning, ES can be effective for optimization problems and more robust than advanced policy gradient methods.
Principles of Mathematical Economics applied to a Physical-Stores Retail Busi...Ashwin Rao
Ashwin Rao presents mathematical models for inventory control and other optimization problems in retail. He describes classical inventory models like economic order quantity and the newsvendor problem. These problems involve stochastic optimization and can be modeled as Markov decision processes. Rao also discusses multi-item and multi-location inventory control problems faced in real-world retail and how they require solving large MDPs using techniques like dynamic programming and reinforcement learning.
Understanding Dynamic Programming through Bellman OperatorsAshwin Rao
Policy Iteration and Value Iteration algorithms are best understood by viewing them from the lens of Bellman Policy Operator and Bellman Optimality Operator
A.I. for Dynamic Decisioning under Uncertainty (for real-world problems in Re...Ashwin Rao
The document discusses artificial intelligence techniques for dynamic decision making under uncertainty and provides examples from retail and financial trading. It introduces the framework of stochastic control, which involves optimization over time with uncertain and evolving variables. The key problems discussed are inventory control in retail and portfolio optimization in finance. Both are characterized as Markov decision processes with states, actions, rewards, and transitions between states governed by probabilities. The document outlines solutions to basic single-period versions and then more complex multi-period versions of these problems, highlighting the challenges of large and complex state and action spaces as well as delayed or unknown consequences of actions.
Overview of Stochastic Calculus FoundationsAshwin Rao
1) Sample paths of Brownian motion are continuous but almost always non-differentiable. They are of infinite total variation but finite quadratic variation.
2) Ito's lemma and Ito isometry relate stochastic integrals of Brownian motion to integrals of deterministic functions, allowing stochastic processes to be analyzed using tools from calculus and probability theory.
3) The Fokker-Planck and Kolmogorov equations link stochastic differential equations to partial differential equations governing the probability density function of the process, while the Feynman-Kac formula relates certain PDEs to conditional expectations of the process.
Risk-Aversion, Risk-Premium and Utility TheoryAshwin Rao
This lecture helps understand the concepts of Risk-Aversion and Risk-Premium viewed from the lens of Utility Theory. These are foundational economic concepts used widely in Financial applications - Portfolio problems and Pricing problems, to name a couple.
To make Reinforcement Learning Algorithms work in the real-world, one has to get around (what Sutton calls) the "deadly triad": the combination of bootstrapping, function approximation and off-policy evaluation. The first step here is to understand Value Function Vector Space/Geometry and then make one's way into Gradient TD Algorithms (a big breakthrough to overcome the "deadly triad").
Stanford CME 241 - Reinforcement Learning for Stochastic Control Problems in ...Ashwin Rao
This document provides an overview of a course on reinforcement learning for stochastic control problems in finance. The course will cover the theory of Markov decision processes and algorithms like dynamic programming and reinforcement learning. It will apply these algorithms to three financial problems: dynamic asset allocation, optimal exercise of American options, and optimal trade order execution. The course aims to blend theory, programming, and modeling of real-world finance problems. It will involve coding assignments and cover foundational papers on applying reinforcement learning in finance domains.
HJB Equation and Merton's Portfolio ProblemAshwin Rao
Deriving the solution to Merton's Portfolio Problem (Optimal Asset Allocation and Consumption) using the elegant formulation of Hamilton-Jacobi-Bellman equation.
The document summarizes the policy gradient theorem, which provides a way to perform policy improvement in reinforcement learning using gradient ascent on the expected returns with respect to the policy parameters. It begins by motivating policy gradients as a way to do policy improvement when the action space is large or continuous. It then defines the necessary notation, expected returns objective function, and discounted state visitation measure. The main part of the document proves the policy gradient theorem, which expresses the policy gradient as an expectation over the discounted state visitation measure and action-value function. It notes that in practice the action-value function must be estimated, and proves the compatible function approximation theorem, which ensures the policy gradient is computed correctly when using an estimated action-value
A Quick and Terse Introduction to Efficient Frontier MathematicsAshwin Rao
A Quick and Terse Introduction to Efficient Frontier Mathematics. Only a basic background in Linear Algebra, Probability and Optimization is expected to cover this material and gain a reasonable understanding of this topic within one hour.
Towards Improved Pricing and Hedging of Agency Mortgage-backed SecuritiesAshwin Rao
Ashwin Rao presented a seminar on improving pricing and hedging of agency mortgage-backed securities. The current industry standard approach uses an option-adjusted spread to account for prepayment risk, but this has challenges. Rao proposed an alternative approach that calibrates and trades based on the market price of prepayment risk rather than a constant option-adjusted spread. This involves modeling sources of prepayment risk other than interest rates, such as forecast errors in prepayment models, and deriving risk-neutral processes for these factors using their market prices of risk.
Recursive Formulation of Gradient in a Dense Feed-Forward Deep Neural NetworkAshwin Rao
Recursive Formulation of Gradient in a Dense Feed-Forward Deep Neural Network. Derived for a fairly general setting where the supervisory variable has a conditional probability density modeled as an arbitrary Generalized Linear Model's "normal-form" probability density, and whose output layer activation function is the GLM canonical link function.
The document discusses how estimating regional demand correlations more accurately can provide significant inventory reduction benefits. It shows that as correlations decrease from 20% to 10% to 0%, the total safety stock across warehouses decreases non-linearly, dropping by approximately 28% and 78% respectively. With deeper risk pooling by doubling the number of warehouses from 20 to 10, inventory reductions increase further, with an 84% drop if correlations go from 20% to 0%. Accurately estimating lower correlations allows companies to take greater advantage of these non-linear inventory benefits.
The document describes an omni-channel newsvendor problem where a single inventory level must satisfy demand across multiple populations or channels. It considers the case of two channels/populations (A and B) with a given joint demand distribution. The problem is to determine the minimum common inventory level (I) and allocation fraction (γ) to satisfy required service levels for each population, given an allocation rule for how shortages are distributed when demand exceeds inventory. Analytical solutions are developed for the two-channel case and extensions to more than two channels are noted.
This document discusses the mathematical similarities between call/put option pricing in derivatives trading and the newsvendor problem in supply chain optimization.
The key points are:
1) Call/put option pricing and the newsvendor problem can both be formulated as expectations of "hockey stick" payoff functions, with the newsvendor problem equivalent to optimizing a portfolio of calls and puts.
2) Under certain assumptions like a Gaussian distribution, the formulas for call/put prices and newsvendor costs are analogous and involve concepts like delta hedging.
3) In both cases, when the strike/supply is optimized, the cost becomes insensitive to changes in the underlying stock price/expected demand,
The Fuss about || Haskell | Scala | F# ||Ashwin Rao
The document discusses functional programming languages Haskell, Scala, and F# and their advantages over imperative programming. It notes they encourage a different problem-solving approach using concepts like immutable values, recursion, and avoiding side effects. Specific features highlighted include static typing with type inference, algebraic data types, pattern matching, and monads for controlling program flow and errors in a functional way. Examples show equivalent algorithms in different languages to demonstrate these concepts.
Careers outside Academia - USC Computer Science Masters and Ph.D. StudentsAshwin Rao
Talk given at USC CS Colloquium to grad students (http://viterbi.usc.edu/news/events/?event=10265). The topic was - Prospective Careers outside Academia.
IIT Bombay - First Steps to a Successful CareerAshwin Rao
Career-Advice for IIT Bombay students. Emphasis on Quant Finance Jobs. Decisions-making on Tech versus Finance, Startups versus Large Firms. Introduction to ZLemma.com to help students identify the right jobs/careers.
Stanford FinMath - Careers in Quant FinanceAshwin Rao
This document provides an overview of careers in quantitative finance from the perspective of Ashwin Rao, who has worked as a VP of quant strategies at Goldman Sachs and managing director of modeling at Morgan Stanley. It defines quantitative finance as roles requiring advanced math, stats, and computer science skills. Examples of roles include derivatives traders, strategists, modelers, and algorithmic trading quants. The document offers advice for students, such as developing coding skills, and warns that much learning will happen on the job. It also describes evaluating candidates' abilities during interviews and the company ZLemma's tools for algorithmic career guidance.
Success is often not achievable without facing and overcoming obstacles along the way. To reach our goals and achieve success, it is important to understand and resolve the obstacles that come in our way.
In this article, we will discuss the various obstacles that hinder success, strategies to overcome them, and examples of individuals who have successfully surmounted their obstacles.
Learnings from Successful Jobs SearchersBruce Bennett
Are you interested to know what actions help in a job search? This webinar is the summary of several individuals who discussed their job search journey for others to follow. You will learn there are common actions that helped them succeed in their quest for gainful employment.
A.I. for Dynamic Decisioning under Uncertainty (for real-world problems in Re...Ashwin Rao
The document discusses artificial intelligence techniques for dynamic decision making under uncertainty and provides examples from retail and financial trading. It introduces the framework of stochastic control, which involves optimization over time with uncertain and evolving variables. The key problems discussed are inventory control in retail and portfolio optimization in finance. Both are characterized as Markov decision processes with states, actions, rewards, and transitions between states governed by probabilities. The document outlines solutions to basic single-period versions and then more complex multi-period versions of these problems, highlighting the challenges of large and complex state and action spaces as well as delayed or unknown consequences of actions.
Overview of Stochastic Calculus FoundationsAshwin Rao
1) Sample paths of Brownian motion are continuous but almost always non-differentiable. They are of infinite total variation but finite quadratic variation.
2) Ito's lemma and Ito isometry relate stochastic integrals of Brownian motion to integrals of deterministic functions, allowing stochastic processes to be analyzed using tools from calculus and probability theory.
3) The Fokker-Planck and Kolmogorov equations link stochastic differential equations to partial differential equations governing the probability density function of the process, while the Feynman-Kac formula relates certain PDEs to conditional expectations of the process.
Risk-Aversion, Risk-Premium and Utility TheoryAshwin Rao
This lecture helps understand the concepts of Risk-Aversion and Risk-Premium viewed from the lens of Utility Theory. These are foundational economic concepts used widely in Financial applications - Portfolio problems and Pricing problems, to name a couple.
To make Reinforcement Learning Algorithms work in the real-world, one has to get around (what Sutton calls) the "deadly triad": the combination of bootstrapping, function approximation and off-policy evaluation. The first step here is to understand Value Function Vector Space/Geometry and then make one's way into Gradient TD Algorithms (a big breakthrough to overcome the "deadly triad").
Stanford CME 241 - Reinforcement Learning for Stochastic Control Problems in ...Ashwin Rao
This document provides an overview of a course on reinforcement learning for stochastic control problems in finance. The course will cover the theory of Markov decision processes and algorithms like dynamic programming and reinforcement learning. It will apply these algorithms to three financial problems: dynamic asset allocation, optimal exercise of American options, and optimal trade order execution. The course aims to blend theory, programming, and modeling of real-world finance problems. It will involve coding assignments and cover foundational papers on applying reinforcement learning in finance domains.
HJB Equation and Merton's Portfolio ProblemAshwin Rao
Deriving the solution to Merton's Portfolio Problem (Optimal Asset Allocation and Consumption) using the elegant formulation of Hamilton-Jacobi-Bellman equation.
The document summarizes the policy gradient theorem, which provides a way to perform policy improvement in reinforcement learning using gradient ascent on the expected returns with respect to the policy parameters. It begins by motivating policy gradients as a way to do policy improvement when the action space is large or continuous. It then defines the necessary notation, expected returns objective function, and discounted state visitation measure. The main part of the document proves the policy gradient theorem, which expresses the policy gradient as an expectation over the discounted state visitation measure and action-value function. It notes that in practice the action-value function must be estimated, and proves the compatible function approximation theorem, which ensures the policy gradient is computed correctly when using an estimated action-value
A Quick and Terse Introduction to Efficient Frontier MathematicsAshwin Rao
A Quick and Terse Introduction to Efficient Frontier Mathematics. Only a basic background in Linear Algebra, Probability and Optimization is expected to cover this material and gain a reasonable understanding of this topic within one hour.
Towards Improved Pricing and Hedging of Agency Mortgage-backed SecuritiesAshwin Rao
Ashwin Rao presented a seminar on improving pricing and hedging of agency mortgage-backed securities. The current industry standard approach uses an option-adjusted spread to account for prepayment risk, but this has challenges. Rao proposed an alternative approach that calibrates and trades based on the market price of prepayment risk rather than a constant option-adjusted spread. This involves modeling sources of prepayment risk other than interest rates, such as forecast errors in prepayment models, and deriving risk-neutral processes for these factors using their market prices of risk.
Recursive Formulation of Gradient in a Dense Feed-Forward Deep Neural NetworkAshwin Rao
Recursive Formulation of Gradient in a Dense Feed-Forward Deep Neural Network. Derived for a fairly general setting where the supervisory variable has a conditional probability density modeled as an arbitrary Generalized Linear Model's "normal-form" probability density, and whose output layer activation function is the GLM canonical link function.
The document discusses how estimating regional demand correlations more accurately can provide significant inventory reduction benefits. It shows that as correlations decrease from 20% to 10% to 0%, the total safety stock across warehouses decreases non-linearly, dropping by approximately 28% and 78% respectively. With deeper risk pooling by doubling the number of warehouses from 20 to 10, inventory reductions increase further, with an 84% drop if correlations go from 20% to 0%. Accurately estimating lower correlations allows companies to take greater advantage of these non-linear inventory benefits.
The document describes an omni-channel newsvendor problem where a single inventory level must satisfy demand across multiple populations or channels. It considers the case of two channels/populations (A and B) with a given joint demand distribution. The problem is to determine the minimum common inventory level (I) and allocation fraction (γ) to satisfy required service levels for each population, given an allocation rule for how shortages are distributed when demand exceeds inventory. Analytical solutions are developed for the two-channel case and extensions to more than two channels are noted.
This document discusses the mathematical similarities between call/put option pricing in derivatives trading and the newsvendor problem in supply chain optimization.
The key points are:
1) Call/put option pricing and the newsvendor problem can both be formulated as expectations of "hockey stick" payoff functions, with the newsvendor problem equivalent to optimizing a portfolio of calls and puts.
2) Under certain assumptions like a Gaussian distribution, the formulas for call/put prices and newsvendor costs are analogous and involve concepts like delta hedging.
3) In both cases, when the strike/supply is optimized, the cost becomes insensitive to changes in the underlying stock price/expected demand,
The Fuss about || Haskell | Scala | F# ||Ashwin Rao
The document discusses functional programming languages Haskell, Scala, and F# and their advantages over imperative programming. It notes they encourage a different problem-solving approach using concepts like immutable values, recursion, and avoiding side effects. Specific features highlighted include static typing with type inference, algebraic data types, pattern matching, and monads for controlling program flow and errors in a functional way. Examples show equivalent algorithms in different languages to demonstrate these concepts.
Careers outside Academia - USC Computer Science Masters and Ph.D. StudentsAshwin Rao
Talk given at USC CS Colloquium to grad students (http://viterbi.usc.edu/news/events/?event=10265). The topic was - Prospective Careers outside Academia.
IIT Bombay - First Steps to a Successful CareerAshwin Rao
Career-Advice for IIT Bombay students. Emphasis on Quant Finance Jobs. Decisions-making on Tech versus Finance, Startups versus Large Firms. Introduction to ZLemma.com to help students identify the right jobs/careers.
Stanford FinMath - Careers in Quant FinanceAshwin Rao
This document provides an overview of careers in quantitative finance from the perspective of Ashwin Rao, who has worked as a VP of quant strategies at Goldman Sachs and managing director of modeling at Morgan Stanley. It defines quantitative finance as roles requiring advanced math, stats, and computer science skills. Examples of roles include derivatives traders, strategists, modelers, and algorithmic trading quants. The document offers advice for students, such as developing coding skills, and warns that much learning will happen on the job. It also describes evaluating candidates' abilities during interviews and the company ZLemma's tools for algorithmic career guidance.
Success is often not achievable without facing and overcoming obstacles along the way. To reach our goals and achieve success, it is important to understand and resolve the obstacles that come in our way.
In this article, we will discuss the various obstacles that hinder success, strategies to overcome them, and examples of individuals who have successfully surmounted their obstacles.
Learnings from Successful Jobs SearchersBruce Bennett
Are you interested to know what actions help in a job search? This webinar is the summary of several individuals who discussed their job search journey for others to follow. You will learn there are common actions that helped them succeed in their quest for gainful employment.
Leadership Ambassador club Adventist modulekakomaeric00
Aims to equip people who aspire to become leaders with good qualities,and with Christian values and morals as per Biblical teachings.The you who aspire to be leaders should first read and understand what the ambassador module for leadership says about leadership and marry that to what the bible says.Christians sh
In the intricate tapestry of life, connections serve as the vibrant threads that weave together opportunities, experiences, and growth. Whether in personal or professional spheres, the ability to forge meaningful connections opens doors to a multitude of possibilities, propelling individuals toward success and fulfillment.
Eirini is an HR professional with strong passion for technology and semiconductors industry in particular. She started her career as a software recruiter in 2012, and developed an interest for business development, talent enablement and innovation which later got her setting up the concept of Software Community Management in ASML, and to Developer Relations today. She holds a bachelor degree in Lifelong Learning and an MBA specialised in Strategic Human Resources Management. She is a world citizen, having grown up in Greece, she studied and kickstarted her career in The Netherlands and can currently be found in Santa Clara, CA.
Joyce M Sullivan, Founder & CEO of SocMediaFin, Inc. shares her "Five Questions - The Story of You", "Reflections - What Matters to You?" and "The Three Circle Exercise" to guide those evaluating what their next move may be in their careers.
We recently hosted the much-anticipated Community Skill Builders Workshop during our June online meeting. This event was a culmination of six months of listening to your feedback and crafting solutions to better support your PMI journey. Here’s a look back at what happened and the exciting developments that emerged from our collaborative efforts.
A Gathering of Minds
We were thrilled to see a diverse group of attendees, including local certified PMI trainers and both new and experienced members eager to contribute their perspectives. The workshop was structured into three dynamic discussion sessions, each led by our dedicated membership advocates.
Key Takeaways and Future Directions
The insights and feedback gathered from these discussions were invaluable. Here are some of the key takeaways and the steps we are taking to address them:
• Enhanced Resource Accessibility: We are working on a new, user-friendly resource page that will make it easier for members to access training materials and real-world application guides.
• Structured Mentorship Program: Plans are underway to launch a mentorship program that will connect members with experienced professionals for guidance and support.
• Increased Networking Opportunities: Expect to see more frequent and varied networking events, both virtual and in-person, to help you build connections and foster a sense of community.
Moving Forward
We are committed to turning your feedback into actionable solutions that enhance your PMI journey. This workshop was just the beginning. By actively participating and sharing your experiences, you have helped shape the future of our Chapter’s offerings.
Thank you to everyone who attended and contributed to the success of the Community Skill Builders Workshop. Your engagement and enthusiasm are what make our Chapter strong and vibrant. Stay tuned for updates on the new initiatives and opportunities to get involved. Together, we are building a community that supports and empowers each other on our PMI journeys.
Stay connected, stay engaged, and let’s continue to grow together!
About PMI Silver Spring Chapter
We are a branch of the Project Management Institute. We offer a platform for project management professionals in Silver Spring, MD, and the DC/Baltimore metro area. Monthly meetings facilitate networking, knowledge sharing, and professional development. For more, visit pmissc.org.
IIT-B 2012 Pre-Placements Career Guidance Talk - Transcript
1. * Life After IIT
: IIT tag follows you. Is it a good thing ? IIT stereotypes - smart, arrogant
: "We were so naive during our IIT days !" is a statement I have heard from
several IIT alumni. At IIT, one interacts with a homogeneous set of people
leading to a herd mentality on campus. We get convinced that our worldview is
correct when it is often grossly wrong. My own beliefs at IIT -  Programming
should be avoided like the plague  - look what happened to me. I've launched a
tech startup and really enjoy coding now.
: Acquire versatility in skills and Personality. Balance hard and soft skills,
similarly balance your personality between aggressive and charming
: Associate with people different than you. Make friends with non-IITians and
people from different cultural backgrounds. My 5 years in LA were the most
significant years of my life because I got tremendous exposure to a wide range
of people
* What are suitable jobs for me ?
: You can figure it out objectively based on the set of credentials you have
acquired so far. Should not be based on whims. Use hard data to figure this out
: Glamour, Money and the Mom Sydrome. 3 big mistakes people make - they chase
jobs that sound cool, they chase jobs that pay the most, they chase jobs that
makes people close to them happy - your mom or your girlfriend or to get the
respect of your friends
: Understand your Skills and your Taste. Determine areas where you have the
skills and where you also have the taste, i.e. strong interest. They are not
always correlated. My own example: I am not exceptionally gifted at Math, but
Math makes me very happy. So Math academia is not for me. I chose the
intersection of finance, math and CS - I am good at this intersection and the
intersection has also been a happy space for me.
: Think Career, not Jobs. Look forward 5-10 years - Would you like to be a
prof ? Would you like to a trader ? Would you like to be a corporate advisor ?
Would you like to be a marketer ? Would you like to be an engineer ?
* Prepare well, but chill out
: Research the various jobs thoroughly. Read up on the web. Read some books
: Advice from classmates and seniors ? Be careful - Herd mentality on campus and
potentially biased advice from immediate seniors. Seek out people a few years
senior, and especially those who don t have a personal agenda when they give
advice.
: Personal life versus Professional life. Chill out: Getting a great job is not
a matter of life and death. Personal life is way more important than
professional life.
: Don't ever compare yourself with others ! Peer pressure seems to be the source
of most stress
: Easy to correct bad jobs or wrong decisions. Even if you start with not such a
great job, you will have ample opportunity to correct
* Higher Studies
: Titles like MBA or PhD - what is their value ? We like to collect titles
because it makes us feel important. The reasons are often shallow.
: What do the best employers care about ? The best hiring groups generally
don t care about advanced degrees. They mainly want people from good undergrad
schools. There are a few jobs which care about Ph.D. or MBA
: Ph.D. -> only if you see yourself as an academic. Probably the sole reason to
do a Ph.D.: If you see yourself as a lifetime academic or researcher. I enjoyed
my Ph.D. but by the time I was done, I didn t want to remain narrowly
specialized. I wanted to be versatile.
: MBA -> Reset Career. Two reasons to do an MBA: 1) If you are sure you want to
switch your career to something very different, press the career reset button by
doing an MBA. MBA Campus recruiting puts you in front of employers in your new
domain. 2) Another reason is the alumni network. But very important to go to a
2. top B school
: Masters versus Ph.D. I am generally not in favor of IIT undergrad students
doing a Masters. The degree doesn t buy you much as most Masters programs have
coursework at levels below the IIT undergrad courses. One reason to do it: If
you want to do Core Enginering jobs or want to work in another country, doing a
Masters from that country might be valuable
: MBA: Now or later ? From the US or from India ? My advice is to try a job for
a few years and then if you feel you want to do something different, apply to a
top-rated B School. IIMs are useful if you are keen on working in India and on
Indian markets or products.
* Finance Jobs
: Quite a range of options for people with technical backgrounds
: Important to have the right personality
: Algo trading roles are very hot right now
: We still don t have enough exciting finance jobs in India.
* Current Wall Street Scenario
: Wall Street definitely in an unhealthy state right now. Is Wall Street dying ?
No
: Regulations hurting innovation and profitability of Wall Street firms. But
some regulations are good to contain Wall Street excesses and the culture has
sobered down quite a bit
: Compensation levels are still higher than other industries but severely down
from 5 years ago
: Still good for STEM backgrounds. Great future for Quants. Everything is
electronic these days. Hard skills are now valued enormously.
: The tide will turn. I am confident Wall Street will roar back. They still hire
the best people from the best schools.
* STEM Jobs
: Boom in Tech, especially in Silicon Valley. Tech jobs have suddenly become
very attractive
: Tech jobs don't pay well ? Some tech jobs are paying as much or more than
finance jobs
: Big Data has created a phenomenon where people with advanced STEM skills -
like Data Scientists are hot commodities
: Google, Facebook, Apple, Amazon. The 4 firms which have created a revolution
in the tech industry by spawning interesting companies and interesting jobs
within their particular business lines.
: Exciting tech jobs in India ? Some exciting tech jobs in Bangalore, mainly
startups. Big tech firms are still not doing enough cutting-edge work in India
* Start ups
: Not for everyone. Personal reasons for starting ZLemma: I have always been
enamored with doing my own thing but Wall Street has been good to me. Key
reasons for doing it now: I have some financial security now and I sense the
opportunity to do something big with my startup
: Market Size, Product, Team. These are the 3 questions you should answer before
you try your own startup. How large is your market size in terms of customers
and revenue ? Why is your product better than the competition ? Is your team
capable of executing on the vision ?
: Some challenges: How will you get funded ? Do you have a good set of
advisors ? How passionate are you about your idea ?
: It s a lot of fun but also a lot of risk-taking. Can get very stressful.