From idea to launch.
At Ecowell, I worked across frontend, backend, database, and mobile dependencies in a three-person engineering team.
I’m Abhay Sharma. I build full-stack products, developer tools, and resilient systems that are useful at scale.
Currently: B.S. Data Science and Applications · IIT Madras
From product interfaces to infrastructure, I focus on measurable outcomes.
My work combines product judgment with hands-on engineering. These are the strengths I bring to a team.
At Ecowell, I worked across frontend, backend, database, and mobile dependencies in a three-person engineering team.
I helped launch software for 10,000+ customers and 1,000+ daily tickets, and built CF Insights for 20,000+ users.
My projects span offline-first WebSocket systems, rigorous C++ data structures, APIs, analytics, and cross-platform apps.
GitHub figures update from my public profile. The visitor counter is seeded at 100, as requested.
Engineering across web, mobile, data, and operations within a small team.
Ecowell India
Built and launched a customized CRM within a three-person engineering team, supporting 10,000+ customers across 8 locations and coordinating operations across 18 teams.
Products and systems built end to end, with performance and reliability in mind.
Codeforces analytics for 20,000+ users and approximately 200 daily active users.
Analyzes ratings, contests, submissions, solved problems, and topic-wise performance to generate personalized insights and improvement plans. Supported peak usage above 100 API calls per second.
Local-network communication without an internet connection or cloud infrastructure.
A PWA with 7 REST endpoints and 1 WebSocket route for room messaging, member presence, moderation, and CSV exports. Includes a persistent offline queue, 2-second reconnection, service-worker caching, and configurable 30-day retention.
Bounded 200-message history, per-client rate limiting, origin validation, role-protected moderation, and audit logging across 3 SQLite entities; 8 automated tests passed.
Reusable cache with average O(1) lookup, insertion, update, and eviction.
Includes an interactive CLI and thread-safe wrapper. Validated with 100,000 randomized reference-model operations, edge cases, exception checks, and an 8-thread stress test covering 40,000 reads.
Benchmarked on 200,000-operation workloads: up to 57.6× the throughput of a linear vector baseline, with 4 GCC/Clang CI configurations and sanitizer checks.
Field notes on building products, exploring systems, and learning in public. Click a story to read it.
Articles and reading times load with the blog.
Loading articles…
Languages and tools chosen for the problem, from interface details to underlying systems.
C++, Python, JavaScript, TypeScript, SQL, HTML, CSS
React, React Native, Node.js, Express.js, Django, Flask, FastAPI
PostgreSQL, MySQL, SQLite, NumPy, pandas, scikit-learn, Matplotlib, Seaborn
Git, Linux, CMake, GitHub Actions, REST APIs, WebSockets, pytest, Ruff
Bachelor of Science (BS) in Data Science and Applications
Competitive programming and community work that shaped how I solve problems and help others grow.
Advanced to Round 2 of Meta Hacker Cup 2025.
Reached Pupil rank and solved 500+ problems across competitive programming platforms.
Codeforces profile ↗LeetCode profile ↗Founded and led a competitive programming community, organizing regular problem-solving sessions for IIT Madras BS students.
I’m always interested in thoughtful software projects and engineering conversations. Let’s connect at 24f1001915@ds.study.iitm.ac.in.