Building Corvane | AI/ML Engineer | Cloud Native AI & Real Time Systems | Quantitative Finance

Bengaluru, Karnataka, India

Pune City, Maharashtra, India
Hybrid







Thapar Institute of Engineering & Technology
Narayana Junior College - India
Join Parth and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal BossBuilding Corvane | AI/ML Engineer | Cloud Native AI & Real Time Systems | Quantitative Finance
I am an AI/ML & Cloud Engineer who builds LLM systems that run in production and ship real impact. My work spans generative AI, agentic pipelines, multi-LLM orchestration, cloud backends, and privacy-preserving infrastructure, with a focus on FinTech, renewable energy, and e-commerce. I am also building toward quantitative finance and systematic trading, with probability, market microstructure, and low-latency C++ systems as my focus. My starting point is Monte Carlo risk modelling and strategy backtesting. As an AI Intern at Iris, I shipped production systems for real users: - Built an agentic template-generation pipeline with memoised workflow decomposition, cutting project setup latency by 70% and automating 100% of new-user onboarding - Engineered a multi-LLM orchestration layer routing across Claude, GPT-4, and Gemini with capability-based dispatch and dynamic context management - Traced a critical billing bug to an off-by-one error in the credits ledger and restored the pipeline within 24 hours At 42Labs, I built a biometric tokenization pipeline that converts face embeddings into irreversible 256-bit tokens, so zero raw embeddings are ever stored. I also merged a fix into Apple's FoundationDB, a production-grade distributed database. Projects: - Solari Scope: cloud fleet orchestration that fans jobs out across isolated microVMs, with LLM-generated worker scripts, SSE result streaming, guaranteed sandbox teardown, and live cost alerts - CreditIQ: AI credit intelligence platform with a deep learning classifier (82.5% AUC-ROC), a ChromaDB RAG pipeline, and FinBERT sentiment scoring (87% accuracy), fully Dockerized - LiveRisk: Monte Carlo risk engine running 10,000 correlated paths with LSTM forecasting and 95% VaR/CVaR, built on FastAPI, Postgres, and Redis Recognition: Top 10% (#193/2,118) at QuantStorm 2026, Top 15 at the SBI Hackathon (Global Fintech Fest 2026), and Top 10 Finalist at the AI for Sustainability Hackathon 2026. 3rd year CSE @ TIET · Open to AI/ML, Software & Cloud Engineering Summer Internships for 2027. Also exploring quant developer and quant research roles in trading and fintech. Email: parthmd018@gmail.com
Greater Bengaluru Area
claude code
claude
cursor
gemini
langchain
langfuse
Co-Founder & CEO at Corvane, Oct 2026 - Present · 2 mos, Bengaluru, Karnataka, India, Building Corvane in stealth, leading customer discovery, go-to-market and company direction for an early-stage venture in the Software and AI space. - Driving outreach to potential customers, running discovery conversations, and turning early interest into the first design partners. - Setting strategy and priorities for the team, and representing the venture to mentors, incubators and investors, while keeping product details confidential. - Leading the design of Corvane's agentic AI workflows, the systems that read, check and draft for each shipment while keeping a person in charge of every decision.
Technical Intern at FORTYTWO LABS, Aug 2026 - Present · 4 mos, Pune City, Maharashtra, India
AI Engineer at IrisVision, Mar 2026 - May 2026 · 2 mo, Hybrid, -Designed and deployed an LLM-powered agentic template generation pipeline with automated reasoning loops and memoised workflow decomposition, cutting project setup latency by 70% and eliminating manual configuration for 100% of new user onboarding flows - Built multi-LLM orchestration layer routing across Claude, GPT-4, and Gemini with capability-based dispatch, dynamic context management, and intelligent load balancing serving real production users - Diagnosed and resolved a critical billing infrastructure bug by tracing root cause through payment-flow logs; identified off-by-one error in credits ledger, patched and regression-tested, restoring full monetisation pipeline within 24 hours
Marketing Member at Entrepreneurship Development Cell, TIET, Aug 2025 - Dec 2025 · 5 mos
Thapar Institute of Engineering & Technology
Narayana Junior College - India
github: https://github.com/PARTHDESHMUKH2005