Hi, my name is
Pranav Avva.
Senior Software Engineer at Hadrius
Princeton CS grad building scalable systems at the intersection of data, engineering, and automation. Full-stack web, applied ML, and cloud infrastructure.
// about
I'm a Senior Software Engineer at Hadrius (YC W23), where I build scalable systems powering AI-driven financial and regulatory compliance. My work spans full-stack development, distributed infrastructure, and the intersection of data and automation.
Before Hadrius, I helped grow BoreDM from its early stages into a production-scale SaaS platform for geotechnical data management — cutting API costs by 77%, slashing page load times by 60%, and architecting infrastructure for SOC 2 compliance.
I studied Computer Science with a minor in Finance at Princeton, where I conducted ML research at the Andlinger Center for Energy and Environment, served as a teaching assistant for core CS courses, and led tech for The Daily Princetonian.
Outside of work, I'm an FAA Part 107-licensed drone pilot, a home lab enthusiast running self-hosted services on Docker, and usually either strength training or out on a trail somewhere.
Education
Princeton University
B.S.E. Computer Science
Minor in Finance
2020 — 2024
Location
New York, NY
// experience
Senior Software Engineer
Hadrius YC W23
Oct 2025 — Present
New York, NY
- Building scalable systems powering AI-driven financial and regulatory compliance
- Full-stack development, distributed infrastructure, and data automation pipelines
Software Engineer
May 2022 — Oct 2025
Phoenix, AZ
- Designed and shipped full-stack features (React, Flask, PostgreSQL) for a geotech data SaaS platform used by national engineering firms
- Reduced third-party API costs by 77% ($19k annually) through backend caching optimization
- Cut initial page load time by ~60% by streamlining frontend dependencies
- Built internal tools and Slack integrations for cross-team cohesion between dev, sales, and ops
- Architected AWS VPC/EC2 infrastructure aligned with SOC 2 Type II requirements
Teaching Assistant
Princeton CS Dept.
Jan 2022 — Dec 2022
Princeton, NJ
- Supported 3 core CS courses (COS 126, 217, 226): data structures, memory management, algorithm design
- Led office hours and debugging sessions; mentored 200+ undergrads
Chief Technology Officer
The Daily Princetonian
Aug 2021 — Sep 2022
Princeton, NJ
- Led digital infrastructure for a 400+ member student media organization
- Built monetized web tools increasing unique site traffic by 45% YoY
- Modernized legacy systems to scalable backend infrastructure
ML Research Intern
Andlinger Center for Energy and Environment
Jun 2022 — Aug 2022
Princeton, NJ
- Predicted magnetic core loss using a Seq2Seq LSTM neural network (>95% test accuracy)
- Presented findings at the Center's 2022 Annual Meeting
// projects
LSTM-based regression model predicting next-day stock prices using 10 years of market data and sentiment from 6 million Reddit posts. Demonstrated that retail investor sentiment improves predictive power, particularly for high-cap firms.
CNN and U-Net models for automatic grayscale image colorization, trained on ImageNet-1K in CIELAB color space. U-Net achieved 25-point lower test loss with better object-boundary coherence. Trained on Princeton's Adroit HPC with NVIDIA V100 GPUs.
Guess the Elo
Chess rating prediction from single-game analysis. Preprocessed 31,000+ Lichess games with Stockfish evaluations, engineered domain features (centipawn loss, time controls, openings). Best model achieved RMSE of 159 rating points.
Expense Ledger
Personal finance tool for tracking HSA/FSA qualified medical expenses and transit/parking commuter benefit accounts. Helps maximize tax-free fringe benefit reimbursements.
Home Lab
Self-hosted Docker services including media server, automated backups, and home automation. Configured VPN and firewall rules for secure remote access from untrusted networks.