ISHITA SAHU
01. Architecture & Identity

INTELLIGENT
SYSTEMS &
HARDWARE.

B.Tech Electronics & Communication Engineering student at IIIT Kottayam. Specializing in hardware-software co-design, Generative AI multi-agent workflows, FPGA acceleration, and real-time embedded ADAS architectures.

Ishita Sahu
LOC: BENGALURU, IN
NAME: ISHITA SAHU
DEGREE: B.TECH ECE (2024-2028)
AT IIIT KOTTAYAM
PRIMARY DOMAIN: EMBEDDED & GEN-AI
CONTACT: +91 8123550181
02. Industry & Research Deployment

PROFESSIONAL EXPERIENCE

FILTERED VIEW: ALL SYSTEMS
[TRACK: EMBEDDED AI & HARDWARE]
May 2026 — Present | Bengaluru, India

PROJECT TRAINEE — ADAS SOFTWARE

Transportation Business Unit (Tata Elxsi)
  • Developed 1,000+ lines of production-grade Python and C++ code for a Maruti Suzuki Surround View System (SVS), implementing interactive gesture-controlled camera navigation using PyQt5, OpenGL, and embedded Linux.
  • Designed a custom path-aware gesture interpolation algorithm converting continuous touch/mouse input into seamless transitions across 8 predefined 6-DOF camera poses without abrupt view switching.
  • Engineered a low-latency frontend-backend communication pipeline using a fixed 98-byte UDP protocol (24 floating-point values + control bytes), sustaining ~20 pose updates/second.
  • Architected a JSON-driven camera configuration framework replacing hard-coded definitions for scalable deployment across Qualcomm and Texas Instruments embedded targets.
[TRACK: DUAL DOMAIN / EDGE AI]
May 2026 — Present | Kottayam, India

RESEARCH INTERN — FPGA ML

Sphinx Lab, IIIT Kottayam
  • Developing FPGA-accelerated ANN and CNN inference pipelines using hls4ml for deployment on Xilinx ZCU102 & Zynq Processing Systems through custom AXI-interfaced IP cores.
  • Implementing quantization-aware neural network architectures and hardware-software co-design techniques to minimize inference latency for Edge AI platforms.
  • Evaluating model latency, resource utilization (LUTs, DSPs), and deployment trade-offs using Vivado, MATLAB/Simulink, and embedded Linux environments.
03. Engineered Systems & Codebases

FEATURED REPOSITORIES

[AI & GEN-AI TRACK]

SCOUT AI

Ishita-2210/scout-ai

Modular five-agent AI assistant built with Google's Agent Development Kit (ADK) and Gemini API. Integrates SQL-based knowledge retrieval with Serper web search in a complex RAG pipeline to eliminate hallucinations and coordinate multi-step tool calling.

Google ADK Gemini API RAG / SQL
[AI & ML TRACK]

TEAMFORGE ENGINE

Ishita-2210/teamforge-recommender

Hybrid talent-to-team recommendation engine combining semantic embeddings, Node2Vec graph embeddings, and XGBoost Learning-to-Rank models. Outperformed semantic baseline by improving Average Precision by 405% (ROC-AUC 0.904).

XGBoost Node2Vec Transformers
[FULL STACK & DATA]

BLINKGRID DASHBOARD

Ishita-2210/blinkgrid-analytics-dashboard

Customer support analytics widget built with React, Express, and PostgreSQL. Engineered optimized SQL aggregation queries and REST APIs to process backend pipelines and drive responsive, low-latency business visualizations.

React / TS Express.js PostgreSQL
[EMBEDDED BCI / ONGOING] ACTIVE

EEG CHESS INTERFACE

Hardware-Software Co-Design

Developing a Brain-Computer Interface (BCI) enabling hands-free chess interaction through EEG neural decoding. Implemented an FBCCA-based SSVEP classification pipeline with a custom ESP32-driven LED visual stimulation module.

ESP32 FBCCA / EEG Signal Processing
[EDGE VISION / ONGOING] ACTIVE

SMART CCTV TRAFFIC

Edge AI & Computer Vision

AI-powered urban traffic management system deploying YOLO and OpenCV to estimate vehicle density from live surveillance streams. Dynamically optimizes four-way signal timings without requiring physical roadside sensors.

YOLO / OpenCV Edge AI Low-Latency
[SIMULATION & GIS]

NANOFLOW SIMULATION

Ishita-2210/nanoparicle_simulation

Microplastic dispersion simulation platform modeling pollutant transport using depth-aware environmental simulation and geospatial datasets. Implemented interactive Folium-based mapping modules to identify environmental hotspots.

Python / Flask Folium Geospatial
[ALGORITHMS & SEARCH]

MINI-GOOGLE SEARCH

Ishita-2210/Mini-Google-Search

Lightweight text-based search engine engineered from scratch. Implements core document tokenization, inverted indexing structures, and efficient ranking algorithms for fast, precise text retrieval and lookup.

Python Indexing Algorithms
[IITG CERTIFICATION]

IIT GUWAHATI ML ARCHITECTURE

Ishita-2210/Capstone_iitg

Architected end-to-end predictive models using supervised learning, CNNs, and ANNs for complex datasets. Features rigorous feature engineering, hyperparameter tuning, and k-fold cross-validation pipelines optimized for real-world deployment.

Supervised Learning CNNs / ANNs Model Optimization
04. Capabilities & Specifications

TECHNICAL SKILLS MATRIX

SYSTEM VERIFICATION: FULL-STACK & HARDWARE READY
[01] PROGRAMMING

Python, C++, C, Verilog HDL, MATLAB, SQL, Java, HTML/CSS, TypeScript.

[02] EMBEDDED & HW

FPGA, hls4ml, Xilinx ZCU102 / Zynq, AXI Interfaces, Embedded Linux, ESP32, ADAS, UDP Networking, Hardware-Software Co-design, Vivado.

[03] GEN-AI & LLMS

Google Gemini API, Google Agent Development Kit (ADK), Multi-Agent Systems, RAG Pipelines, Tool/Function Calling, Prompt Engineering, Serper API.

[04] ML & FULL-STACK

PyTorch, TensorFlow, Scikit-learn, XGBoost, YOLO, OpenCV, OpenGL, React, Express.js, Flask, PostgreSQL, Git, Linux, PyQt5, CMake, Blender.

LEADERSHIP & POSITIONS OF RESPONSIBILITY: Core Member, Beta Labs (AI/ML) | Placement Cell Coordinator | TEDx IIIT Kottayam PR & Outreach Team
05. Academic Foundation

EDUCATION

2024 — 2028 | Kottayam, India

INDIAN INSTITUTE OF INFORMATION TECHNOLOGY KOTTAYAM

B.Tech in Electronics and Communication Engineering
Academic Metric: CGPA: 8.23 / 10.0
06. Professional Verification

CERTIFICATIONS & HONORS

LINKEDIN LEARNING — FPGA DEVELOPMENT HARDWARE / EMBEDDED

Completed comprehensive hands-on training covering FPGA architecture fundamentals, digital logic design, and hardware implementation workflows using Field-Programmable Gate Arrays.

MACHINE LEARNING CERTIFICATION — IIT GUWAHATI TOP 10% RANKING

Ranked among the top 10% of participants in an advanced certification program covering supervised learning, deep neural networks, and intelligent systems design.

AICTE ATAL FDP — CHIP-LEVEL SYSTEMS (CUSAT) SECURE HARDWARE

Completed an intensive Faculty Development Program on Secure and Intelligent Chip-Level Systems with specialized emphasis on embedded hardware architectures and secure AI computing.

ISRO HACKATHON 2025 SPACE APPS / CV

Engineered a Python-based computer vision pipeline for automated lunar crater detection from satellite imagery at a national-level AI and Space Applications hackathon.

GOOGLE DEVELOPERS VIBE CODING INTENSIVE COHORT

Selected for an intensive 5-day cohort-based program focusing on cutting-edge AI-assisted software development practices and generative tools.