AI Systems Engineer | Backend Architect | Embedded Product Builder
π B.Tech CSE (Artificial Intelligence) β VIT Pune (2023β2027)
π CGPA: 8.52 | Last Semester SGPA: 9.38
I design and ship production-grade systems β from embedded firmware to scalable AI-powered cloud platforms.
Backend Systems β’ AI Infrastructure β’ Embedded Engineering
Architected and shipped the complete digital infrastructure for an early-stage startup as a Freelance Software Engineer.
- π Zero-to-One Architecture: Built the startup's entire digital presence from scratch. Architected a multi-tenant platform featuring Flutter mobile apps, a Next.js admin dashboard, and a Dockerized Spring Boot backend exposing 340+ production APIs (Dev β Staging β Prod workflow).
- π€ AI Infrastructure: Developed a GenAI content engine using automated AI crawlers and a LangChain video generator to create promotional media for service providers. Integrated a Vet AI assistant for symptom-based pet health insights.
- π Enterprise Security: Implemented a robust auth system using JWT with refresh token rotation, BCrypt-12, JTI revocation blacklisting, RBAC, and secure AWS S3 presigned upload pipelines.
- π Startup Impact: Successfully launched the product to market, bootstrapping 40+ active app users, generating 300+ initial website visitors, improving SEO from 0 to ~700 daily impressions, and orchestrating 50+ automated transactional email workflows.
π Letter of Appreciation: View Letter
π View Backend Performance Metrics (JMeter)
| Virtual Users | Requests | Avg Response | P95 Latency | Throughput | Error Rate |
|---|---|---|---|---|---|
| 1,000 | 10,000 | ~81 ms | ~100 ms | ~99 req/sec | ~0.2% |
Performance optimizations include HTTP caching, ETag validation, and optimized DB queries.
π Pune | Oct 2024 β Dec 2024
- ESP32 Shuttle Launcher: Built a closed-loop RPM control system (Embedded C/C++) achieving a 1-second launch cycle. Reduced manufacturing costs by 80%, piloted with All-India ranked players, and generated 50+ pre-orders.
- Offline OTP Generator: Implemented RFC-6238 TOTP authentication, deploying secure offline auth for 200+ users.
Stack: FastAPI β’ LLMs β’ Python (asyncio) β’ gTTS
- Multi-Agent Pipeline: Decomposed unpredictable LLM generation into a strict, 6-stage structured pipeline (Discovery, Culinary Intelligence, Ranking, Clustering, Priority, Routing).
- Multilingual & Audio Intelligence: Engineered a native-script-enforced narration system with concurrent Text-to-Speech (TTS) generation for multi-language travel guides.
- Production-Grade Reliability: Implemented strict JSON schema validation, deterministic scoring outside the LLM, intelligent caching, and parallel async execution to drastically reduce latency.
Stack: FastAPI β’ React β’ MongoDB β’ Twilio β’ LiveKit β’ Deepgram
- Engineered a real-time AI voice agent capable of conversational voice calls, automated meeting scheduling, full transcript generation, and a live monitoring dashboard.


