Jan 2023 – Apr 2023 · 3 mos

• Directed global quantitative R&D across clearing, risk, and trading systems, leading C++/Python quantitative library engineering and ML model pipelines for derivatives and market infrastructure. Managed model risk governance for clearing houses, delivering regulatory-compliant frameworks that underpin ICE’s global risk, margin, and settlement engines.
• Built and scaled core C++ math and ML libraries for stochastic calculus, PDE solvers, Monte Carlo, and optimization, enabling risk teams to deploy low-latency, production-grade models across equities, rates, and commodities. Established engineering standards for distributed computing, concurrency, and SIMD-MPI optimized HPC systems, ensuring scalable performance at petabyte data scale.
• Led the design of quant model reporting dashboards for clearing, with flexible filters by model type, asset class, timeframe, and participant, providing real-time transparency to regulators and clearing participants. Advanced ICE’s infrastructure by integrating HPC, grid computing, and OpenCL accelerated ML pipelines into production, bridging quant research with enterprise-grade deployment.
• Mentored and scaled India-based R&D teams to global leadership standards, driving initiatives from mathematical model innovation → library integration → production rollout. Delivered full life-cycle engineering for clearing models: research, validation, stress testing, reporting, and supervisory audit alignment.
• Positioned as a quantitative engineering leader, combining deep C++ systems expertise with ML innovation to enhance ICE’s role as the world’s premier market infrastructure provider.

Jan 2021 – Jan 2023 · 2 yrs

• Advised by University Of Warwick,
• University Of Amsterdam,
• Indian Institute Of Science,
• Courant Institute Of Mathematical Sciences,
• Oxford MAN Institute Of Quantitative Research,
• Leaders from Nomura Research, Wells Fargo & Bloomberg.
• continuous IMA Champions,
• and
• Pierre and Marie Curie University
• Analytics & Risk Platforms
• 1. Market Data & Ingestion
• Aggregated crypto derivative exchange APIs.
• Normalized order book, trade, and vol surface data.
• Built data enrichment layer: forward curves, composite vol indices.
• 2. Analytics & Quant Models
• Developed options pricing engines (Black–Scholes, local vol, stochastic vol).
• Produced implied vol surfaces, greeks, and sensitivity measures.
• Built risk decomposition models for portfolio-level exposure.
• 3. Low-Latency & HPC Analytics
• Designed HPC pipelines for real-time surface recalibration.
• Implemented grid/distributed compute for large-scale backtesting.
• Optimized analytics delivery via streaming APIs.
• 4. APIs & Client Access
• Built enterprise APIs for analytics delivery (REST + streaming).
• Integrated dashboard UI for traders & analysts.
• Supported institutional integration into existing risk systems.
• 5. Research & Thought Leadership
• Published peer-reviewed crypto derivatives research.
• Delivered quant commentary on volatility, risk, and pricing.
• Embedded research into analytics platform delivery.
• 6. Governance & Reliability
• Ensured high-availability analytics clusters.
• Delivered SLA-driven reliability for institutional clients.
• Built audit/compliance reporting (crypto regulatory frameworks).

Aug 2020 – Dec 2020 · 4 mos

• Transportation Modeller (Nonlinear C++ Engine)
• (emphasis: math + algorithms + C++ engine design)
• Architected and implemented linear optimization engines (LP-based supply–demand balancing, cost minimization) and extended them into nonlinear optimization engines for congestion-aware routing, equilibrium assignment, and time-dependent logistics modeling.
• Designed and implemented solver algorithms in C++ (custom simplex, interior-point, gradient-based, and iterative nonlinear solvers) optimized for numerical stability, performance, and scalability.
• Modeled nonlinear cost structures such as congestion delay functions, nonlinear fuel consumption, emissions, and stochastic delays, enabling realistic representation of transportation physics.
• Built core C++ optimization libraries as reusable components, serving as the mathematical backbone of Blue Yonder’s Transportation Management solutions.

Apr 2018 – Aug 2020 · 2 yrs 4 mos

• Transportation Planner (TMS – C++ Business Objects Engine)
• (emphasis: business logic + integration + planning workflows)
• Integrated linear and nonlinear optimization cores into the Transportation Management System (TMS) as reusable C++ business object engines, abstracting mathematical complexity for planners and schedulers.
• Designed business workflows around transportation planning: carrier assignment, multi-modal routing, congestion-aware scheduling, and equilibrium-based load distribution.
• Enabled real-time and batch planning modes, supporting both deterministic supply–demand flows and dynamic, congestion-sensitive routing decisions.
• Delivered measurable business impact by improving routing efficiency, cost reduction, and adaptability of large-scale supply chain planning under nonlinear, real-world constraints.
• Ensured enterprise-grade integration with Blue Yonder’s supply chain platform, making the optimization engines accessible through APIs, user interfaces, and planning dashboards.

Nov 2015 – Apr 2018 · 2 yrs 5 mos

• Led AI/ML engineering for Oculus headset platforms, focusing on generative AI, automatic speech recognition (ASR), natural language processing (NLP), and computer vision (CV). Directed cross-functional teams developing real-time multimodal inference pipelines, optimized for GPU hardware on-device.
• Architected GPU-accelerated AI kernels and low-latency model execution frameworks in C++ and Python, integrating transformer-based models with ONNX Runtime and custom GPU optimizations. Designed and deployed quantized inference pipelines to meet power and latency constraints critical for immersive AR/VR.
• Delivered production-ready speech-to-text, conversational AI, and vision-based scene understanding, enabling natural human-computer interaction in headset environments. Partnered with research scientists and product engineers to transition state-of-the-art ML models (transformers, multimodal fusion architectures) into scalable deployment on Oculus devices.
• Built and mentored AI/ML engineering groups, embedding best practices in GPU optimization, multithreading, concurrency, and model lifecycle management. Enhanced end-user experience by ensuring real-time responsiveness of AI-driven voice, vision, and contextual reasoning features inside Oculus Reality Labs products.

Sep 2013 – Jul 2015 · 1 yr 10 mos

• 1. HSBC FX Algos / AES (Algorithmic Execution Suite)
• Purpose: Algorithmic FX execution with smart order routing.
• Key Components / Workflows:
• Pre-Trade TCA → cost/impact modelling, liquidity heatmaps.
• Execution Strategies → TWAP, VWAP, Get Done, Implementation Shortfall, Liquidity Seeking, PoV.
• Smart Order Router (SOR) → low-latency venue selection, order slicing.
• Post-Trade TCA → performance analytics, benchmarking against market VWAP/MID.
• Connectivity → FIX/API, direct market access.
• 2. HSBC Low-Latency Equities & Messaging Infrastructure (Internal Engineering)
• Purpose: High-performance compute + low-latency messaging for equities/FX.
• Key Components / Workflows:
• Messaging Fabric → ultra-low-latency bus, affinity/NUMA-aware.
• Compute Layer → NUMA-optimized multi-core servers, FPGA/GPU acceleration (select workloads).
• Market Data Handlers → direct feeds (co-location, FPGA tick-to-trade).
• Order Router → deterministic latency for smart order routing.
• Resiliency Layer → failover, replay, disaster recovery.
• Cool Things I Did,
• Analyze market microstructures, cost of trade, and liquidity.
• Write execution algos to optimize order based strategies. (VWAP, TWAP, and more)
• Use ML, stats modeling, and various optimization techniques to improve execution
• Backtest and simulate your system to identify performance bottlenecks and opportunities for improvement
• Partner with traders, developers, and leadership to get sh*t done.
• Things Used,
• Market Microstructure and Algo Trading fundamentals
• TradFi markets
• Coding in Python, C++
• Stats modeling and ML algos, and associated optimization techniques.
• Order routing and management and market connection protocols

Jun 2012 – Sep 2013 · 1 yr 3 mos

• Architected and scaled distributed storage and content delivery systems powering FiOS TV Video-on-Demand (VOD) and Subscription VOD (SVOD) platforms, supporting millions of concurrent users across the U.S. Designed low-latency, high-throughput pipelines for media ingest, transcoding, storage, and edge delivery, ensuring seamless streaming at petabyte scale.
• Engineered clustered storage solutions with data replication, sharding, and distributed caching to guarantee high availability and fault tolerance. Led development of multithreaded C++ and Java services optimized for I/O-intensive workloads, achieving sub-second content retrieval across tens of thousands of nodes.
• Built real-time monitoring, capacity planning, and load-balancing frameworks, enabling proactive scaling during peak demand events (e.g., sports finals, premieres). Collaborated with network engineering teams to integrate content distribution networks (CDNs) and adaptive bitrate streaming, optimizing end-user experience across diverse devices and bandwidth conditions.
• Delivered next-gen SVOD infrastructure, enabling Verizon to compete in the early OTT streaming landscape against Netflix and Hulu. Contributed to patent-grade innovations in storage tiering, metadata indexing, and distributed video delivery, cementing FiOS TV as a market leader in large-scale digital media systems.
• This was a tight, technical, and leadership-oriented, role in distributed systems, storage scaling, and video delivery at petabyte scale

Aug 2007 – Oct 2010 · 3 yrs 2 mos

• Core Engineering: Contributed to low-level algorithm development and optimization for experimental Kinect integrations and embedded system pipelines.
• Kinect Depth Camera: Developed and optimized C++ drivers and middleware for Kinect 3D camera, focusing on depth sensing, ASR/NLP, and CV pipelines.
• Embedded Systems / WinCE: Worked on algorithm portability and optimization for WinCE-based platforms, ensuring interoperability with Microsoft’s embedded OS stack.
• Algorithm Optimization: Refactored and accelerated real-time sensor fusion, computer vision, and audio recognition pipelines for low-latency interaction.
• Systems Performance: Improved concurrency, scheduling, and resource utilization in constrained embedded environments, delivering reliable performance under tight compute/memory budgets.
• Cross-Disciplinary Work: Collaborated with software and hardware teams on C++ CV/ASR workloads, pioneering embedded ML/CV pipelines before GPU acceleration was standard.