Skip to content
View ilambharathim's full-sized avatar

Block or report ilambharathim

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
ilambharathim/README.md
Ilam Bharathi — Deep Space Galaxy Circuit Header



Engineering Telemetry Dashboard



Galaxy Divider



Silicon & Architecture Domains



Galaxy Divider

✨ Featured Projects


⚡ RIVA — RTL Intelligent Verification Assistant

SystemVerilog • Logic Simulation • Waveform Analysis • SVA Assertions • ● ACTIVE

  • Developing an automated verification framework combining SystemVerilog testbenches, simulation regressions, waveform analysis, and linting rules for specification-driven verification closure.
  • Traces functional discrepancies from testbench assertions directly to failing RTL datapath signals.
  • Repository: github.com/ilambharathim/AI_RTL_ASSISTANT
RIVA Waveform Timing Simulation Trace


🚀 SNN-Based Object Detection — Microchip PolarFire SoC

Spiking Neural Networks • PolarFire SoC FPGA • RISC-V Coprocessor • Edge AI Acceleration

  • Exploring hardware-oriented deployment of Spiking Neural Networks (SNN) on the PolarFire SoC FPGA platform, optimizing spike-timing dynamics and weight quantization for resource-constrained edge vision.
  • Exploits PolarFire non-volatile flash architecture for near-zero static leakage power while eliminating redundant multiply-accumulate operations during static frames via event-driven spike sparsity.

PolarFire SoC FPGA Hardware Subsystem: ┌────────────────────────┐ AXI4 Bus ┌────────────────────────────────┐ │ 4x 64-bit RISC-V Cores │ ◀────────────────▶ │ Neuromorphic SNN Accelerator │ │ (Supervisory Control) │ │ • Leaky Integrate & Fire (LIF) │ └────────────────────────┘ │ • Event-Driven Spike Sparsity │ └────────────────────────────────┘



🔶 22 nm 6T FinFET SRAM Cell — Area Scaling & Layout

22nm FinFET • 6T SRAM • KLayout • SEMulator3D • Synopsys Custom Compiler

  • Designed and analysed a 6-Transistor (6T) FinFET SRAM memory cell to study fin pitch scaling, Static Noise Margins (SNM), and 3D physical process profiles.
  • Fin Pitch: 30 nm with minimized parasitic capacitance between adjacent channel fins.
  • Read SNM: > 180 mV ensuring non-destructive read operations under process variation.
  • Hold SNM: > 280 mV with stable data retention down to {DD,min} = 0.65 ext{V}$.


🟢 180 nm CMOS Capacitance-to-Digital Converter

Cadence Virtuoso • Spectre • 180nm Bulk CMOS • Switched-Capacitor PVT

  • Transistor-level design and simulation of an analog switched-capacitor CDC with comprehensive AC, transient, and parasitic evaluation across -40°C to 125°C PVT corners.
  • Sensitivity: < 0.15% deviation across ±10% {DD}$ supply variations.
  • Resolution: Monotonic 10-bit digital representation of micro-capacitive sensor inputs.


🔷 4.8 GHz PLL Analog & Mixed-Signal Verification

Cadence Virtuoso • Spectre • PFD / Charge Pump / VCO • Phase Noise • VERIFIED

  • Verified a 4.8 GHz Phase-Locked Loop subsystem validating lock range, settling behavior, jitter performance, and open-loop stability.
  • Lock Range: 4.4 GHz to 5.2 GHz with fast settling response (< 1.8 µs).
  • Phase Noise: -112 dBc/Hz offset at 1 MHz with 58° open-loop phase margin across all corners.


🔴 Gesture-to-Speech FPGA System

Verilog HDL • AMD Xilinx Vivado • Synchronous FSM • Real-Time DSP

  • Offline FPGA hardware translating analog flex-sensor kinematic profiles into synthesized audio indices using a low-latency Verilog FSM with deterministic < 15 ms response.


✨ Semiconductor Image Restoration — SEMICON India 2026

PyTorch 2.0 • NAFNet • SEM Metrology • 35.10 dB PSNR • KLA Track PS01

Metric Target Baseline TEAM KIRAH NAFNet Gain
Peak SNR (PSNR) 29.40 dB 35.10 dB +5.70 dB Gain
Structural Similarity (SSIM) 0.9200 0.9885 High Edge Fidelity
Inference Latency (H100) 12.0 ms < 2.4 ms Real-Time Metrology Speed

Galaxy Divider

🧰 Technical Toolchain

SystemVerilog Verilog VHDL Cadence Virtuoso Synopsys VCS

Vivado PolarFire SoC KLayout SEMulator3D Python



💼 Experience & Research

` May 2026 – Jun 2026 Research Intern // IIITDM Kancheepuram • Hardware-efficient AI accelerator architectures & edge quantization.

Oct 2025 – Nov 2025 Project Intern // Synopsys Centre of Excellence (CIT) • 180 nm CMOS Capacitance-to-Digital Converter in Cadence Virtuoso & Spectre.

May 2025 – Jun 2025 Embedded Systems Intern // Phoenix Soft Tech • Microcontroller bus interfacing (SPI/I2C/UART) & peripheral firmware. `


Galaxy Divider

📐 Silicon Implementation Flow

Hardware Engineering Pipeline


🎓 Education & Honors

  • Chennai Institute of Technology — Bachelor of Electronics and Communication Engineering (2024 – 2028)
  • DVCon India 2026 — Contributing to AI-assisted hardware verification & design analysis.
  • Central India Hackathon — Top 7 Finalist (100+ national teams, automated air-quality response).
  • SEMICON India Hackathon 2026 — Team Leader (Team KIRAH), SOTA SEM image restoration.

Galaxy Divider

📬 Connect

Email GitHub

Popular repositories Loading

  1. AI_RTL_ASSISTANT AI_RTL_ASSISTANT Public

    AI-powered SystemVerilog Debugging, Verification, Simulation, Specification Analysis and Auto-Correction Framework.

  2. TEAM-KIRAH_KLA_PSO1 TEAM-KIRAH_KLA_PSO1 Public

    A single-stage, end-to-end deep learning restoration pipeline that takes noisy, downsampled NoisyLR images (128×128128×128) and simultaneously performs speckle denoising, Gaussian deblurring, and l…

    Python

  3. ilambharathim ilambharathim Public

  4. 6t-sram-finfet-area-scaling 6t-sram-finfet-area-scaling Public

    Layout-level area reduction of a 22 nm 6T FinFET SRAM bitcell using coordinated lateral geometric compaction, verified with KLayout and Coventor SEMulator3D

    Python

  5. cdc-mems-microphone-readout cdc-mems-microphone-readout Public

    Time-encoded capacitance-to-digital converter (CDC) with switched-capacitor feedback for bias-free MEMS microphone and microfluidic IDE sensor readout - 180 nm CMOS, Cadence Virtuoso

  6. pll-5ghz-cmos-integer-n pll-5ghz-cmos-integer-n Public

    Transistor-level integer-N PLL targeting ~5 GHz in 90 nm CMOS (GPDK090) - PFD, charge pump, ring VCO, and /48 TSPC divider, simulated in Cadence Spectre