Shivang Raval | Quant Trader & AI Engineer

Shivang Raval

Quant Trader & AI Engineer

Quant Developer | ML & MLOps

Python • C++ • LangChain • PyTorch • Kubernetes

MS CS @ Northeastern

Building high-performance trading systems and production ML infrastructure — from microsecond execution engines to distributed GPU training platforms

About Me

I'm a Quant Trader and AI Engineer currently pursuing my Master's in Computer Science at Northeastern University (4.0 GPA), specializing in quantitative finance and production machine learning systems.

As a Quant Developer, I build trading infrastructure that operates at microsecond precision — from statistical arbitrage strategies achieving Sharpe ratios of 10.5 to low-latency market simulators processing 5.67M events per second.

On the ML & MLOps side, I architect production systems that scale — recently leading a team of 20 engineers at Webearl AI, where we built inference platforms serving 100K+ daily requests with sub-500ms latency. I've built RAG systems with sub-10ms cached response times and distributed training platforms achieving 10.6× speedup.

My tech stack spans the full spectrum: Python and C++ for core systems, PyTorch for ML modeling, LangChain for LLM applications, and Kubernetes for orchestration. I'm passionate about the intersection where quantitative rigor meets engineering excellence.

Featured Projects

From quantitative trading strategies to production ML infrastructure

ML-Driven Statistical Arbitrage

Ensemble trading strategy achieving Sharpe 10.5 through GPU-accelerated feature engineering processing 5M ticks with LightGBM ensemble and hedge ratio optimization

📈 Sharpe: 10.5 | 86% Win Rate | 54K ticks/sec

LightGBMNumbaNumPyPandas

Code

Low-Latency Market Simulator

Multi-venue trading simulator processing 5.67M events/second with microsecond scheduling and 94% better slippage prediction across 8 exchanges

📈 5.67M events/sec | 0.10μs Latency | 94% Slippage Improvement

PythonNumbaSortedContainersML

Code

Production RAG System with RLHF

Retrieval-augmented generation with FAISS vector search achieving sub-10ms cached responses, 100% relevance scores, and human feedback loops

📈 Sub-10ms Cached | 100% Relevance | 60% Hit Rate

LangChainFAISSFastAPIGeminiRedis

Code

Distributed ML Training Platform

Multi-GPU infrastructure achieving 10.6× training speedup (8h → 45min) through data parallelism with demonstrated linear scaling

📈 10.6× Speedup | 2.95× Measured | Linear Scaling

PyTorchDDPKubernetesMLflow

Code

Smart Order Router

Adaptive order router achieving 74% average fill improvement through real-time latency tracking across 8 exchanges, excelling with 87% improvement under network stress

📈 74% Avg Improvement | 87% Under Stress | 8 Venues

PythonReal-timeAdaptive AlgorithmsEMA

Code

High-Throughput Streaming Pipeline

Distributed streaming pipeline achieving 333K+ events/second with <2s end-to-end latency using Kafka and async Python horizontal scaling

📈 333K events/sec | <2s Latency | 5-Node Distributed

KafkaPython AsyncIOPydanticPrometheus

Code

Skills & Expertise

Quantitative Trading

Statistical ArbitrageMarket MicrostructureExecution AlgorithmsKalman FiltersBacktesting

ML & AI Engineering

Production MLDistributed TrainingTime SeriesNLP & RAGModel Optimization

MLOps & DevOps

KubernetesDockerMLflowCI/CDModel Serving

High-Performance Systems

Low-Latency C++ConcurrencyReal-Time ProcessingPerformance Optimization

Core Technologies

PythonC++RustPyTorchTensorFlowLangChainKubernetesDockerKafkaPostgreSQL

Let's Build Something

Whether you're looking for quantitative trading expertise, ML engineering, or technical leadership — I'd love to hear from you.

📍 Boston, MA

Currently: MS CS @ Northeastern University (4.0 GPA)

Open to: Quantitative Trading • ML Engineering • MLOps • Research Opportunities