KESAVANAND M
Hoboken,NJ
Summary
Principal C++ Software Engineer and Applied Quantitative Developer with 18 years of experience architecting ultra-low-latency electronic trading infrastructure, real-time risk engines, and statistical market data pipelines.
Expert in bridging the gap between quantitative research and bare-metal execution using modern C++23, x86_64 microarchitecture optimization, and kernel-bypass networking.
Possess deep applied statistical fluency—holding two patents in probabilistic anomaly detection and natively implementing stochastic mathematical models, jump-diffusion volatility estimators, and high-frequency alpha features directly on the execution hot path to accelerate alpha generation.
Overview
19
19
years of professional experience
Work History
Senior Software Engineer (Quant C++ Developer)
A.R.T Advisors
03.2025 - Current
- Zero-Cost Execution Infrastructure (Currently in Live Production): Designed and engineered an end-to-end modern C++20 infrastructure from the ground up, authoring all core abstractions and test frameworks to bridge real-time normalized global tick data (US, EU, JP, SK, AU equities and futures) with low-frequency models.
- Ultra-Low-Latency IPC & Market Data: Engineered ultra-low-latency decoders for high-frequency normalized feeds, achieving a 5x throughput increase. Built a custom shared-memory snapshot mechanism utilizing x86_64 TSO-optimized Seqlocks and optional futex conditional wakeups to transfer 2D feature tensors directly to quant models for zero-copy statistical inference.
- Bare-Metal Microarchitecture Optimization: Diagnosed and resolved critical x86_64 (Skylake) CPU backend bottlenecks using hardware performance counters. Eliminated ld_blocks_partial.address_alias penalties caused by store-forwarding failures (128-bit vector stores followed by narrow 64-bit and boolean loads in a columnar layout decoder).
- Applied Quantitative Engineering: Bridged quant research and execution by engineering deterministic pipelines to calculate complex statistical features natively on the hot path. Implemented high-frequency bi-power variance, mathematically rigorous return decay signals, buy/sell/retail marked volume classifications etc. directly on the execution hot path. Built comprehensive custom sequencers to construct consolidated books, gracefully managing backwards-flowing time during exact deterministic system replays.
Principal Engineer, Algorithmic Trading
Morgan Stanley
01.2015 - 02.2025
- Kernel-Bypass & Vendor Displacement: Engineered a drop-less packet capture and protocol decoding engine deployed on Solarflare and Napatech NICs (processing ~2M pkts/sec on two cores). Outperformed and fully displaced Corvil vendor solutions, saving millions in annual licensing costs.
- Equities & Options Order Flow Decoders: Architected zero-copy exchange protocol decoders consuming inbound execution reports directly from the kernel-bypass layer. Built highly deterministic parsers for complex, proprietary binary order flow protocols across Equities and Options Market Making on the extreme hot path.
- Real-Time Quantitative Inferencing Engine (Patented): Lead Architect for a firm-wide, ultra-low-latency C++ anomaly detection engine. Designed and implemented the research-to-production pipeline: calculating 100-200 day empirical distributions in Greenplum, persisting symbol-specific parameters to MongoDB, and loading them natively into the C++ hot path for live ECOD (Empirical Cumulative Distribution-based Outlier Detection) scoring. (Awarded 2 Patents as Lead Inventor, including US11360442B2).
- Stochastic Feature Engineering: Engineered statistical estimators directly in C++, including real-time price jump detectors utilizing Lee-Mykland measures. Implemented bipower variance estimators, dynamically optimizing the return frequency based on asset liquidity to mathematically filter out market microstructure noise. Developed dynamic dollar-exposure estimators by calculating the probability of execution using the CDF of the asset's Geometric Brownian Motion.
- Zero-Allocation Order State Management: Designed an in-memory, zero-allocation reference database and generic order correlation framework. Utilized graph algorithms to deterministically link out-of-order execution message chains decoded from the exchange gateways.
Electronic Trading Infrastructure
Morgan Stanley
01.2007 - 01.2014
- Core Infrastructure: Built and maintained latency-sensitive C++ libraries powering the global Electronic Trading Infrastructure.
- Kernel-Bypass Integration: Developed the firm's proprietary low-latency FixSession library, incorporating OpenOnload kernel bypass, SIMD-based checksums, and persistent atomic counters, which remains in critical production use today.
- Compile-Time Optimization: Co-designed "Binary Fix" (internal low-latency messaging format), utilizing C++ template metaprogramming (MPL) to build zero-cost compile-time lookup APIs.
Education
B.Tech - Computer Science
National Institute of Technology
Warangal, India01-2007
Skills
- Languages & Standards: Modern C (C20/C23), Python, SQL (Greenplum, PostgreSQL), MongoDB
- Software Architecture & Design: Zero-Cost Abstractions, Large-Scale Codebase Architecture, Deterministic Testability Frameworks, Highly Iterative Software Correctness, Template Metaprogramming (MPL)
- Systems & Microarchitecture: Ultra-Low Latency Execution, Lock-Free Concurrency, SIMD, x86_64 CPU Microarchitecture Optimization & Performance Analysis
- Quantitative Engineering: Applied Statistics, Stochastic Modeling, High-Frequency Feature Estimators (Bipower Variance, Return Decay), Jump-Diffusion Volatility Models, ML Feature Generation
- Domain: Equities, Futures and Options Market Data, Trading Risk & Execution, Real-time Packet Capture, Timestamp and Exchange Protocol Decoding
- Infrastructure & Tools: Linux Kernel, Networking & Kernel Bypass, OS Tuning, Git, GCC 13, Bazel, RHEL 9
Timeline
Senior Software Engineer (Quant C++ Developer)
A.R.T Advisors
03.2025 - Current
Principal Engineer, Algorithmic Trading
Morgan Stanley
01.2015 - 02.2025
Electronic Trading Infrastructure
Morgan Stanley
01.2007 - 01.2014
B.Tech - Computer Science
National Institute of Technology