Where hardware meets intelligence
Final-year integrated B.Tech/M.Tech student at Jawaharlal Nehru University, specialising in RF & Microwave Engineering — with a parallel depth in multimodal machine learning and local AI systems.
My work spans antenna design in HFSS, clinical NLP pipelines, self-hosted LLM infrastructure, and hardware-level integrations — breadth by design, depth by discipline.
Multimodal PTSD & Depression Detection
Deep learning framework on the DAIC-WOZ dataset. Fine-tuned RoBERTa for clinical transcript classification; extracted 300+ prosodic and spectral features via librosa. Engineered a 7-model audio comparison pipeline (RF, SVM, LR, GBM) utilizing SMOTE and stratified 5-fold CV to counteract severe class imbalance.
OpenSubDL: TUI Subtitle Downloader
High-performance interactive TUI built with Rust to automate subtitle retrieval via the OpenSubtitles API. Engineered a custom regex parsing engine for metadata extraction and cross-compiled deployment for Linux, Windows, and Android using Docker. Optimized binary via Link-Time Optimization (LTO).
Arduino Biometric Data System
Engineered a biometric data acquisition system utilizing an Arduino Uno R4 and an R307S optical sensor. Developed C++ firmware to extract raw 534-byte binary templates, alongside a Python backend to automate reception, detect headers, and save timestamped templates for database integration over a serial interface.
Memristor Non-Volatile Memory Cell
Designed and simulated a 1T-1R memristor-based SRAM alternative in Cadence OrCAD. Characterised emulator circuit across 50–200 kHz; validated hysteresis and retention. Prototyped on a breadboard with oscilloscope validation across temperature extremes (−50°C to +80°C).
2.4 GHz Microstrip Patch Antenna
Rectangular microstrip patch antenna designed and simulated in Ansys HFSS with an FR4 substrate. Applied transmission line theory for effective dielectric constants and utilized an iterative optimization loop to tune resonance frequency and impedance matching.
Text-to-Audio Engine & RAG Assistant
Architected a local RAG assistant for ArchWiki docs using Ollama, ChromaDB, and SentenceTransformers. Parallelly developed an offline TTS web app (Piper, XTTS-v2, Facebook MMS) with dual Docker builds for HuggingFace CPU instances and local NVIDIA-passthrough GPU inference.
Deep stack,
no compromises
AWS Cloud Virtual Intern
- Architected and deployed serverless ML inference pipeline via Lambda, S3, and API Gateway — reduced latency 40% vs. on-premises baseline
- Containerised multi-service application with Docker; orchestrated across AWS EC2 and Google Cloud
- Configured VPC, IAM policies, and security groups with least-privilege access model
Summer Intern — Banaras Locomotive Workshop
- Hands-on training across SCADA systems, Locomotive Test Shop, Telephone Exchange, and Traction Assembly Shop
- Gained working knowledge of locomotive control systems, communication protocols, and safety-critical electronics
- Observed integration of industrial automation and sensor networks in a live production environment
Co-founder — Ephirium Gaming Club
- Organised 5 competitive tournaments (Valorant, CS:GO, CODM) with 150+ cumulative attendees
- Managed end-to-end logistics: registration systems, bracket scheduling, prize distribution
- Collaborated with campus tech clubs to expand the gaming ecosystem
Let's
build
something.
Open to research collaborations, faculty positions, and engineering roles. Based in New Delhi — available globally.