Technical Capabilities & Practice

Engineering, from physical systems to intelligent software.

My engineering interests span semiconductor design, embedded systems, robotics, instrumentation, and software. I approach projects by connecting underlying hardware constraints with system architecture, experimentation, and practical implementation.

Discipline A · Semiconductor Physics

VLSI Design & Low-Voltage SRAM Research

Transistor-level digital schematic capture, sub-1V supply voltage scaling, and characterization of memory cell stability in Cadence Virtuoso.

6T Standard Topology

Shared Access Port Constraints

Cross-coupled CMOS inverters with direct pass-gate access. As VDD scales below 0.8V, voltage divider action during read cycles elevates the storage node voltage (VQ), compromising Read Static Noise Margin (RSNM) and inducing read upsets.

Trade-off: Minimal silicon area vs poor low-voltage read stability.
8T Decoupled Topology

Decoupled Read Buffer Solution

Augments the storage core with an independent 2-transistor read stack. Read discharge is isolated from internal nodes, preserving Read SNM identical to Hold SNM and enabling reliable operation down to near-threshold supply voltages.

Trade-off: ~30% higher layout area for robust sub-threshold noise immunity.
Simulation Workflow Parameters (Cadence Virtuoso)
Simulation Suite
Cadence Spectre
Supply Regimes
1.2V down to 0.6V
Key Metrics
Hold / Read SNM
Process Corners
Nominal & Slow Corners
Cadence Virtuoso Simulation Model · 6T vs 8T Analysis

Low-Voltage SRAM Bitcell Waveform Analyzer

Simulate transient read/write noise margins, access delays, and subthreshold leakage across supply voltages down to near-threshold sub-1.0V regimes.

TRANSIENT WAVEFORM [VDD = 0.80 V]
TIMESCALE: 200 ps/div
TRACES: WL (Wordline) · BL (Bitline) · Q/QB (Storage)CADENCE VIRTUOSO SPECTRE ENGINE
Decoupled Read Path Resilient: 8T topology isolates internal latch storage nodes Q and Q_bar through dedicated read transistors M7-M8, completely eliminating read disturbance down to 0.60V.
Supply Voltage (VDD):0.80 V
0.60 V (Sub-threshold)0.80 V (Low Power)1.20 V (Nominal)
Read SNM:
176 mV
Stable Margin
Write Delay:
75 ps
Access Time
Leakage:
28.3 pA
Subthreshold
8T Memory Cell Architecture:

Decoupled 8T bitcell adds a two-transistor read buffer (M7-M8) driven by read wordline RWL and read bitline RBL, enabling sub-0.8V scaling without sacrificing data stability.

Discipline B · Embedded Firmware & Hardware

Embedded Systems & Telemetry Instrumentation

Interfacing physical transducers, managing low-power sleep modes, parsing serial NMEA protocols, and engineering resilient telemetry links over Sub-GHz LoRa and cellular modems.

Sensory Acquisition

I2C/SPI sensor integration: optical dust counters (PM2.5/PM10), ambient temperature, force-sensing resistors, and ultrasonic transducers with hardware interrupt debouncing.

Wireless Telemetry

Sub-GHz LoRa RF communications for remote off-grid monitoring; dual-band GSM/LTE modems driven via deterministic AT command sequences and checksum verification.

Power Optimization

Hardware-driven interrupt wake-up routines, cyclic sleep scheduling, and quiescent current management enabling extended multi-day battery readiness.

Verified Embedded Implementations
• HerShield Safety Wearable
• Smart Asthma Guardian (UNO R4)
• LoRa Field Telemetry Node
Discipline C · Kinematics & Bionics

Robotics, Articulation & Adaptive Neural Control

Formulating anthropomorphic 5-finger mechanical kinematics, adaptive grasp pressure regulation, and cutaneous haptic feedback loops for prosthetic accessibility.

Official IP MilestoneIssued 27/08/2026

Brain-Controlled Robotic Hand with Adaptive Artificial Intelligence and Haptic Feedback

Design registered and granted by the Patent Office, Government of India, under the Designs Act, 2000 (Design No. 504827-001). Co-inventor listed alongside SRM Institute of Science and Technology.

Adaptive Grasp Regulation

Embedded algorithm evaluating fingertip tactile feedback to detect micro-slip events and automatically modulate servo torque, preventing delicate objects from crushing or dropping.

Cutaneous Haptic Stimulation

Micro-vibrotactile actuators delivering pulse-frequency feedback proportional to surface contact force, re-establishing closed-loop sensory awareness for the wearer.

Discipline D · Scientific Software

Scientific Computing, Simulation & Full-Stack Tooling

Connecting mathematical models with interactive browser interfaces, automated SPICE netlist generators, and high-performance Python data pipelines.

EDA Tooling: RLC Frequency Simulator

Built a complete computer-aided circuit analysis platform in Python, NumPy, Flask, and React. Computes resonance, Quality factor (Q), and generates automated standard SPICE `.cir` netlists with interactive Bode plots.

Security Systems: Mini Firewall Simulator

Designed and implemented a network packet filtering simulator in Python, evaluating incoming/outgoing TCP and UDP streams against 5-tuple security rules with automated access logs.

Engineering Methodology

The 6-Stage Engineering Process

How abstract principles transform into verified schematics, prototypes, and intellectual property.

01

Define the Physical Problem

Isolate fundamental physical, circuit, or operational friction before specifying software abstractions or hardware layouts.

Instance: 6T SRAM bitcell read-disturb upsets during sub-1V VDD scaling.
02

Establish Engineering Constraints

Quantify real boundaries: gate leakage power, supply voltage limits, mass budgets (<450g), signal noise, and real-time latency thresholds (<100ms).

Instance: Bio-potential signal drift and low-voltage inverter noise margins.
03

Synthesize Architecture

Formulate schematic topologies, decoupling buffer nodes, finite-state machine flows, and mathematical transfer kernels.

Instance: Decoupled 2-transistor read buffer in 8T SRAM; closed-loop haptic control loops.
04

Implementation & Netlisting

Draft transistor-level schematics in Cadence Virtuoso, write embedded C++ firmware for microcontrollers, or model vector equations in Python.

Instance: SPICE netlists, predictive models, interrupt service routines, and REST APIs.
05

Empirical Simulation & Evaluation

Execute DC butterfly sweeps, testbench transient analysis, corner simulations (-40°C to +85°C), and prototype bench evaluations.

Instance: Hold/Read SNM extraction in Spectre; tactile slip detection response times.
06

Document Constraints & Future Tapeout

Honestly document parasitic limitations, silicon area penalties (~30% in 8T), and requirements for post-layout verification.

Instance: Disclosing simulation boundaries prior to physical silicon fabrication.

Ready to examine specific project implementations?

Browse the project directory, review the patent certificate, or get in touch.