| Category | Skills & Tech Stack |
|---|---|
| HDL | Verilog SystemVerilog VHDL Chisel |
| Architecture | RISC-V ARM SoC Design AXI/AHB |
| Simulation | Vivado ModelSim Quartus Icarus Verilog |
| EDA | Synopsys DC Cadence Virtuoso OpenROAD Magic VLSI |
| Protocols | PCIe AXI I2C SPI UART JTAG |
| Verification | UVM Formal Verification Cocotb |
| Category | Skills & Tech Stack |
|---|---|
| Languages | Python C++ C JavaScript TypeScript Rust Go |
| Frontend | React Next.js Tailwind CSS TypeScript |
| Backend | Spring Boot FastAPI Node.js Django Flask REST API Apache Maven |
| Databases | PostgreSQL MongoDB Redis SQLite |
| Systems & OS | Linux Bash Assembly Makefile |
| Testing | PyTest Jest Postman |
| Category | Skills & Tools |
|---|---|
| AI / Machine Learning | PyTorch TensorFlow JAX Scikit-learn Hugging Face |
| Deep Learning & GenAI | LangChain LlamaIndex YOLO CLIP |
| Data & Computer Vision | NumPy Pandas OpenCV |
| Software & Web | React Next.js Tailwind CSS FastAPI Django Flask |
| Category | Tools & Infrastructure |
|---|---|
| Cloud Platforms | AWS GCP Azure |
| Containers & Orchestration | Docker Kubernetes |
| CI/CD | GitHub Actions GitLab CI |
| Infrastructure as Code | Terraform Ansible |
| Observability | Prometheus Grafana |
Designed and implemented an low-power neural computation accelerator utilizing bit-serial arithmetic. Won 1st place in Hackathon, NIT Jamshedpur - 2025. (In collaboration with VLSI FOR ALL)
Developed a high-throughput 2D systolic array matrix multiplication engine optimized for CNN workloads. Synthesized and verified on Xilinx FPGAs using Vivado, achieving high resource utilization and optimized clock frequencies.
Developed a high-performance discrete-time voltage-mode PID controller for a synchronous buck converter. Implemented on a Xilinx Spartan-7 FPGA with 23-bit fixed-point precision and parallel hardware execution.
Built a decentralized, mesh-routed offline payment system for zero-connectivity environments. Features store-and-forward gossip protocols and hybrid RSA/AES-GCM encryption for secure offline transactions.
Demonstrated natural language to SQL translation using Oracle AI Database 26ai's Select AI feature. Implemented interactive Jupyter notebooks to automate SQL query generation from plain English prompts.
Developed high-performance custom CUDA kernels for fundamental deep learning operations (SGEMM, 2D convolution, and FlashAttention-style primitives). Optimized memory access patterns via shared memory tiling, coalescing, and register usage to bridge the software-hardware gap.
