Languages, frameworks and tools I use to build production systems.
A sample of what I've shipped recently.
Built a low-latency, full-stack system using C++, Node.js, WebSockets, and Supabase for real-time gym rep counting from accelerometer/gyroscope streams. This project involved defining precise rep-detection logic, streaming sensor data, and developing a responsive HTML/CSS/JS dashboard with integrated computer-vision features for enhanced user experience and live session state tracking.
Developed a Flutter Android ESG-scoring application, which was a Finalist in the SMAC Competition. The app integrated the Gemini API to generate product scores and explanations from barcode/photo/text inputs, and implemented device persistence to save user preferences and history across sessions.
Spearheaded growth initiatives, scaling a student user base from 1,000 to over 17,500+ and significantly growing social media followers to 24,000+. This included leading survey-driven newsletter production, managing an editorial pipeline for 14 student writers, and deploying the organization's first centralized web resource using a modern static site architecture.
A timeline of where I've worked and what I shipped.
Built end-to-end ETL workflows in Python to ingest, clean, and validate global equity and sentiment datasets, transforming raw multi-source inputs into a structured country-month panel for hypothesis-driven time-series and cross-sectional modeling. Performed exploratory data analysis using vectorized rolling correlations, log-differences, and OLS residualization to identify statistically robust lead-lag relationships across markets. Designed and backtested a monthly rebalanced sentiment-driven investment strategy with strict no look-ahead controls, outperforming a global benchmark (CAGR 11.9% vs 9.0%; Sharpe 0.76 vs 0.49; beta 0.73). Led robustness and risk diagnostics in collaboration with a Cambridge professor, formalizing documentation and validation standards to ensure reproducible analytical workflows.
Created a high-throughput automated Python pipeline to process large-scale financial news streams, leveraging MinHash and locality-sensitive hashing (LSH) to eliminate near-duplicate articles at scale before embedding-based clustering. Developed a time-aware semantic clustering framework that dynamically tightens similarity thresholds for temporally proximate articles, isolating true market-moving events from redundant news flow. Developed event-level datasets for downstream research, enabling sentiment extraction and macro signal generation from distinct financial news events rather than raw article-level noise.
Engineered asynchronous, semaphore-controlled API ingestion pipelines in Python (asyncio + httpx) with rate-limit handling, retry logic, and structured validation, reducing total acquisition time by 80% across thousands of equity tickers. Designed concurrent batch processing and Pandas-based transformation workflows to standardize financial metrics across thousands of tickers, using vectorized operations to ensure consistency for downstream analytics and portfolio research. Built indexed MongoDB architectures with upsert refresh logic to enable scalable ETL processing and low-latency queries. Developed a Next.js + TypeScript internal analytics dashboard streaming live financial KPIs, enabling real-time portfolio monitoring, exploratory data analysis, and investment decision support.
Johan Naresh is a versatile and highly analytical professional with expertise spanning Data Science, Quantitative Development, and Data Engineering. He excels at building end-to-end data pipelines, developing sophisticated analytical models, and creating robust software solutions, particularly in financial markets and large-scale data processing. His academic background in Computer Science and Finance, coupled with practical experience, positions him uniquely at the intersection of technology and quantitative analysis.
Throughout his experience, Johan has consistently demonstrated an ability to drive impactful results. He engineered asynchronous API ingestion pipelines that reduced data acquisition time by 80%, designed and backtested an investment strategy outperforming benchmarks with a 11.9% CAGR, and developed automated Python pipelines processing large-scale financial news streams. His work extends to full-stack development, including a real-time analytics dashboard for financial KPIs and a full-stack gym rep counter with computer vision. He has also scaled user bases significantly, growing a student platform from 1,000 to over 17,500 users.
Passionate about leveraging data and advanced algorithms to solve complex problems, Johan is driven by the challenge of transforming raw data into actionable insights and robust, scalable products. He is adept at working with diverse tech stacks and thrives in environments that encourage innovation and a deep dive into technical and quantitative challenges. He is eager to contribute his skills in data-driven strategy, system design, and quantitative modeling to forward-thinking organizations.
Tell me about your project — I'll respond within 24 hours.
We've sent a sign-in link to .
It expires in 15 minutes — check spam if you don't see it.