AI × Finance × Software
Harsh Khandelwal
Building stuff
Engineering student at BITS Pilani who loves solving problems with code. Whether it's building trading algorithms, creating productivity tools, or experimenting with new tech.
About Me
Hey! I'm Harsh, an Electronics and Robotics student who can't stop building things.
What started as curiosity about how technology works has turned into a pretty serious obsession with creating tools that actually solve real problems. I've founded a couple of ventures (MailWave and MarketWire), spent way too many hours debugging trading algorithms, and somehow convinced people to let me lead quantitative research at our finance club.
My sweet spot seems to be where finance meets technology. There's something fascinating about markets—all that data, all those patterns, all those opportunities to build systems that can make sense of the chaos. But I'm not just about the money side of things. I genuinely love the engineering challenges: How do you process thousands of market events per second? How do you make an AI actually understand what someone wants to do with their email? How do you run Python in a browser without everything breaking?
When I'm not coding or analyzing market data, you'll find me tinkering with new frameworks, reading about the latest ML research, or trying to convince my friends that yes, building a browser from scratch is a totally reasonable weekend project.
What I'm Working On
Co-Founder & Lead Developer
2024 - Present
Email is broken. We all know it, but somehow we keep using it for everything. At MailWave, we're trying to fix that by turning your inbox into something that actually works with how you work.
The idea is simple: instead of juggling 15 different tools, you should be able to manage your entire workflow from your email. Need to update a CRM record? Just tell your email to do it. Want to create a Trello card from a client request? Done. Need to pull data from your database? Easy.
What I built:
- Full email infrastructure with IMAP/SMTP integration
- Core AI system that understands natural language commands
- Integrations with Salesforce, Trello, Supabase, DocuSign, and others
- Smart features like auto-summarization and context-aware replies
Early users are saving 5-10 hours per week.
Current Co-ordinator & Algo Labs Head
May 2025 - Present
Leading our quantitative research team, which is a fancy way of saying I get to play with market data and see if we can find patterns that actually make money.
We're working on everything from traditional statistical arbitrage to ML-based prediction models. I'm also exploring some pretty experimental stuff like unikernel-based trading systems (because apparently normal latency optimization isn't enough anymore).
The team has grown to about 15 people, and we're building tools that other clubs and even some external clients are starting to use.
Freelance Developer & Consultant
2023 - Present
My own project—a market intelligence platform that makes sense of all the noise in financial news. It's basically a very smart news reader that tells traders what actually matters.
Built backend systems for their new trading platform. The goal was to compete with Zerodha, so everything had to be fast, reliable, and scalable. No pressure.
Migrated their entire legacy system to something that doesn't crash every Tuesday. Built a unified dashboard that pulls data from various mutual funds and provides risk analytics for wealth managers.
Side Projects & Experiments
Natural Language Trading Strategy Backtester
The idea was simple but technically challenging: what if traders could just describe their strategy in plain English and immediately see if it would make money?
I built a system that takes natural language descriptions like "buy when RSI is below 30 and price is above the 50-day moving average, sell when RSI goes above 70" and converts it into executable Python code using Claude's API.
Technical Implementation:
- Integrated Claude's API for natural language to Python code conversion
- Built a sandboxed WebAssembly-based Python environment in the browser
- Created secure code execution with numpy, pandas, and technical analysis libraries
- Real-time backtesting engine with historical market data integration
- Beautiful performance dashboard showing returns, drawdowns, Sharpe ratios, and risk metrics
Users can now test trading ideas in seconds instead of spending hours coding and debugging.
AI Controller for CST Studio Suite
Working on an AI system that can autonomously design and optimize antennas using CST Studio Suite. Research-stage, but promising results.
Crafts by MailWave
Simple idea: type what kind of email template you want, get a beautiful HTML design. No more fighting with email builders.
Torq Browser
Built a productivity-focused browser using Electron. It tracks your browsing habits and gently reminds you when you're spending too much time on Twitter.
WebAssembly Python Runtime
Figured out how to run scientific Python (NumPy, SciPy) in browsers using WebAssembly. The foundation that made the trading backtester possible.
Technical Stuff
Languages
Python, JavaScript, TypeScript (and whatever else the project needs)
AI/ML
TensorFlow, PyTorch, various NLP libraries. I can train models, but I'm not going to pretend I invented a new architecture.
Web/Backend
Node.js, React, FastAPI, Electron. Standard web stack, nothing fancy.
Data & Infrastructure
PostgreSQL, MongoDB, Redis, Docker, AWS. Can set up CI/CD pipelines without breaking everything.
Education
BITS Pilani (Goa Campus)
B.Tech Electronics and Instrumentation, 2023-2027
Delhi Public School, Patna
2021-2023
Won some tech fest at IIT Patna in 2019, which was pretty cool at the time.
Let's Talk
I'm always up for discussing new projects, sharing war stories about debugging production systems, or just talking about interesting tech problems.
P.S. If you're thinking about building something in fintech or need help with a trading system, definitely reach out. I love this stuff.