Feb 2023 – Present · 3 yrs 1 mo

• Given the increasing unviability of discretionary trading strategies with the changed stock market volatility/price range regime in recent times, we have moved entirely into algorithmic trading.
• Towards this end, we have developed a basket of quant strategies based on functional analysis and stochastic calculus for HFT, and on harvesting time value (theta decay) for more sedate, non-predictive, direction-neutral option writing algorithms.
• The process of trading algorithm development is always ongoing, however, and future scope includes the usage of ML-based deep learning for optimising strategies that are already developed and deployed, as well as for designing and deploying newer, ML-based quant strategies. The current prediction engine for our directional HFT strategy is entirely analytical, with the ML portion used only for trading profit optimisation. Future prediction engines for both HFT and LFT would possibly include, additionally, ML-based algorithms that are currently under development.
• The language of choice is Python used on a Linux development environment. Linux is also the platform of choice for running trading programs on the APIs of the various brokerage firms engaged by us. Professional activities during market hours typically include running rigorously and extensively forward tested, deployed algorithms for revenue generation, and simultaneous collection of curated forward market data for R&D activities related to machine learning and parameter optimisation. Professional activities during non-market hours centre around programming and R&D.

Mar 2013 – Jan 2023 · 9 yrs 10 mos

• Aspire Portfolio Management Pvt. Ltd. is a discretionary fund manager with no directional bias specialised in trading and investments on the National Stock Exchange, India.
• Our major focus is on low frequency, positional discretionary trading strategies based on Elliott Wave analysis, Multiple Time Frame Momentum, Gann timing analysis and correlations with major foreign stock indices. Typical trade holding periods would range between 1 and 6 months. Minor focus going forward (depending on market conditions) will be diversified into medium to long term stock holding, high-frequency program trading and long-term asset investments (private equity and other such).
• Our business model is geared towards client capital growth significantly above the prevalent rate of inflation in India, based on profitability in all market conditions. All aspects of the trade - entry, management and exit are based on a logical and systematic framework of probability that leaves little room for a gambler-like approach.

Jul 2019 – Feb 2020 · 7 mos

• Data architect and enterprise architect in the e-governance initiative of Ernst & Young's Digital Government Advisory division. Currently handling data architecture and enterprise architecture for the establishing an e-governance system for one of the departments in a state government.

Aug 2011 – Feb 2013 · 1 yr 6 mos · Bangalore

• Independent management and business consulting related to problems that require analytics solutions. Delivered the following projects in analytics:
• 1. Built a logistic regression based classifier for fraud detection in automobile insurance claims involving
• two or more parties, for an insurance firm
• 2. Built a sentiment classifier for brand feedback based on the Naïve Bayes Classifier – 73% accuracy with
• training data and 70% accuracy with test data from another brand.
• 3. Built a model to value a delinquent mortgage portfolio of housing loans for a major domestic bank by
• constructing a predictive model for estimating the recovery rate using multivariate linear regression.
• 4. Built a logistic regression based decision model for a major domestic bank for determining whether or
• not an unsecured personal loan should be made out to a new customer, based on relevant inputs in the
• loan application form.
• 5. A multivariate regression based solution, implemented in C++, to the problem of estimating radio reach for a radio station, the purpose of which was to allocate resources to set up radio stations in 250 new cities in India, based on data collected from 25 cities where they had existing radio stations.
• 6. An algorithmic solution to the online search engine based ad budget optimisation problem for a market research firm with clients in the online advertising business.

Jul 2005 – Jul 2011 · 6 yrs · Bangalore, Gurgaon, Mumbai

• Responsible as single resource in the IT Risk Consulting Group (part of the Global Consulting Practice) for wireless security.
• Responsible as a project lead to deliver algorithms for search on a role-based access control application for a large engineering client.
• Delivered a project as lead, the mandate of which was to perform automated testing of a cryptographic product offering of TCS.
• Project Lead in a manual testing project for a BFSI client with continuous delivery commitments.
• Project Lead in a Production Support project for back-end credit cards process for a BFSI client with continuous delivery commitments.
• Designed and implemented an algorithm to solve the basis-change problem for LDPC decoding matrices to reduce cost of multiplication in the decoding process.
• Formulated a methodology for wireless security assessment in consultation with a TCS partner.
• Adapted an existing factoring scheme for linear wavelet transforms into a non-linear setting, particularly in the domain of morphological wavelets.
• Developed algorithms for search on the intranet site of a client.
• Delivered a research proposal on the semantic gap problem in multimedia, in particular with respect to video and image applications.
• Explored the Self-Organizing Map (SOM) methodology in the context of ECG defect classification.

Sep 2003 – Jun 2005 · 1 yr 9 mos · Bangalore

• Created protocols for secure dynamic group key agreement for active nodes tracking a moving target within a wireless sensor network surveillance area, intent tracking for moving targets.
• Building a prototype for ad hoc wireless sensor network for monitoring applications using geographic zone formation and efficient data aggregation and query distribution.
• Part of a team that successfully built a prototype for ad hoc wireless sensor network.
• Implemented pair wise secure key agreement, using an existing scheme based on bivariate symmetric polynomials over GF(2^m), which was then used to build a group key agreement protocol.

Jan 2003 – Jan 2003 · 0 mo

• Research on finding evidence of long-range behaviour in wireless network traffic.