Somak Somanchi
Software Engineer, Quant-Focused
I build and operate production systems — from full-stack platforms to market-data infrastructure and paper-trading engines that run unattended in live markets.
My current focus is quantitative research and trading systems: signal validation, backtesting discipline, and the operational rigor that keeps a strategy honest once it's live.
Backend, infrastructure, and DevOps are the muscle behind that — I care about systems that hold up under real conditions, not just in a demo.
About Me
My Introduction
I work across the stack, but the through-line is systems that have to be correct under real conditions: production infrastructure, market-data pipelines, and the testing, incident-response, and deploy discipline that keeps them trustworthy once they're live.
Quantitative trading is where most of my focus goes right now — building paper-trading engines against live market data, and treating every promising signal as a hypothesis to falsify before it ships, not a result to confirm.
Open to quant-focused software engineering roles and conversations about market systems, trading infrastructure, or backend architecture.
Skills
My Technical Skills
Work
Projects, Experience & Certifications
Algorithmic Trading Research Platform
A multi-market paper trading system I built and actively run: a BTC perpetual futures engine marked against live Delta Exchange data, and an options paper trading engine for NIFTY / SENSEX / FINNIFTY marked against live Groww prices, backed by a research lab for regime classification and signal backtesting. In production long enough to accumulate a real operational history — 25 merged PRs, 560+ tests, and 3 live incidents diagnosed and fixed.
View Full Case Studytrendline-ml : Learning to Draw Trendlines
A human-in-the-loop tool for teaching a model to draw price trendlines the way I would by hand. A Streamlit app lets me scroll through OHLC data and draw trendlines directly on the chart; each one becomes a labeled example — the price window plus the line's two endpoints. Those examples train a model to predict trendline endpoints from price action alone, and its predictions get shown back in the same tool for accept / correct / reject, closing the loop and growing the dataset with every session.
QuickFix : Web App + PWA
Built a cross-platform Progressive Web App (PWA) offering a seamless experience across desktop and mobile. Integrated smart caching, offline access, and installable features to ensure speed, reliability, and near-native usability.
Umbrella Intelligence Platform
A data intelligence platform for unifying, cataloging, and analyzing enterprise data assets. Provides data modeling, ERD documentation, table cataloging, and DDL scripts, with AI integration to support analytics, data governance, and business intelligence workflows.
Job Platform Backend
Backend for an automated job data ingestion and enrichment platform. Integrates multiple data sources, processes job feeds, and enriches listings with AI-powered services, backed by Python, BigQuery, and Upstash, with support for analytics, notifications, and geolocation.
Sociopedia : MERN based social media platform
Developed a full stack social media platform using MERN stack that us fully function and responsive for all device screens.
Krypt Transactions
Devised a 100% secure Ethereum transactions app that uses least gas fuel written in compact and efficient code. Also allows user to express feelings with giphs.
Followers Predictor
Built this model using the Facebook Prophet library is utilized to create a predictive model for future follower counts. It generates forecasts for several months ahead and displays the results in a plot with 92% accuracy.
Evolutionary Classifier
Uses evolutionary algorithm searches for optimal CNN architectures for image classification on CIFAR-10 dataset. Iteratively selecting best-performing ones, applying crossover and mutation, and evaluating their fitness.
PromptPulse : News Research Tool
Designed a news research tool [PromptPulse]: user-provided URLs, article segmentation, OpenAI embeddings, FAISS vector database indexing, semantic search. Optimized with MapReduce for token reduction, achieving a 35% time complexity reduction.
Dashboard Buddy Backend
Backend API service for the Dashboard Buddy application, providing user management, referral tracking, and analytics functionality with robust data management and real-time insights for referral programs.