Tanush Singla

I’m a Computer Science student who builds high-performance systems across ML pipelines, systems programming, and full-stack development. I’m drawn to problems where engineering choices directly translate into measurable improvements in latency, memory, and reliability.

One highlight is my custom 8-bit Scalar Quantization pipeline for a vector search engine. I compressed float32 embeddings by 75% while preserving full semantic search, patched a silent uint8 overflow in NumPy accumulation, and implemented Asymmetric Dot Product search with 0.28ms query latency—validated at 100% Recall@10 against a brute-force baseline.

I also develop end-to-end ML frameworks, including a predictive stock analysis pipeline. I ingest real-time market API data, engineer SMA/RSI/MACD features, benchmark linear regression, random forest, and LSTM models, and keep inference under 15ms by removing redundant NumPy allocations. My backtesting module evaluates predictions over six months of historical data, and my OOP-structured preprocessing reduced execution time by ~40% versus a naive baseline.

On the product side, I ship user-facing web systems and iterate with strong engineering discipline. I built and deployed a fest web app handling 1,000+ concurrent users with zero downtime, cut page asset size by ~30% with image compression and lazy-loading, and delivered modular components with full mobile responsiveness. I’m also actively leading as Founder & CEO on a B2B SaaS MVP for education clients, and I value teams that measure outcomes and move fast without breaking quality.