Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Senior AI/ML Engineer with 10+ years of software experience and 3+ years building production computer vision and ML systems. Expertise spans pose estimation, object detection, multi-object tracking, and end-to-end video inference pipelines deployed on GCP, with healthcare-adjacent applications including medical imaging diagnostics and movement/biomechanical assessment. Strong in full SDLC ownership—from HPC/SLURM training and multi-GPU optimization to real-time user-facing deployment, monitoring, and CI/CD. Also delivers GenAI solutions with LLMs, RAG, and agentic workflows using GPT-4/Gemini/LLaMA and LangChain/LangGraph, and bridges technical outcomes with product, QA, and leadership to drive reliable, measurable impact.…Senior AI/ML Engineer with 10+ years of software experience and 3+ years building production computer vision and ML systems. Expertise spans pose estimation, object detection, multi-object tracking, and end-to-end video inference pipelines deployed on GCP, with healthcare-adjacent applications including medical imaging diagnostics and movement/biomechanical assessment. Strong in full SDLC ownership—from HPC/SLURM training and multi-GPU optimization to real-time user-facing deployment, monitoring, and CI/CD. Also delivers GenAI solutions with LLMs, RAG, and agentic workflows using GPT-4/Gemini/LLaMA and LangChain/LangGraph, and bridges technical outcomes with product, QA, and leadership to drive reliable, measurable impact.

Karthika Ramasamy

AI Engineer, Software Engineer, Data Scientist, +2





Senior AI/ML Engineer with 10+ years of software experience and 3+ years building production computer vision and ML systems. Expertise spans pose estimation, object detection, multi-object tracking, and end-to-end video inference pipelines deployed on GCP, with healthcare-adjacent applications including medical imaging diagnostics and movement/biomechanical assessment. Strong in full SDLC ownership—from HPC/SLURM training and multi-GPU optimization to real-time user-facing deployment, monitoring, and CI/CD. Also delivers GenAI solutions with LLMs, RAG, and agentic workflows using GPT-4/Gemini/LLaMA and LangChain/LangGraph, and bridges technical outcomes with product, QA, and leadership to drive reliable, measurable impact.…Senior AI/ML Engineer with 10+ years of software experience and 3+ years building production computer vision and ML systems. Expertise spans pose estimation, object detection, multi-object tracking, and end-to-end video inference pipelines deployed on GCP, with healthcare-adjacent applications including medical imaging diagnostics and movement/biomechanical assessment. Strong in full SDLC ownership—from HPC/SLURM training and multi-GPU optimization to real-time user-facing deployment, monitoring, and CI/CD. Also delivers GenAI solutions with LLMs, RAG, and agentic workflows using GPT-4/Gemini/LLaMA and LangChain/LangGraph, and bridges technical outcomes with product, QA, and leadership to drive reliable, measurable impact.

Available to hire

Senior AI/ML Engineer with 10+ years of software experience and 3+ years building production computer vision and ML systems. Expertise spans pose estimation, object detection, multi-object tracking, and end-to-end video inference pipelines deployed on GCP, with healthcare-adjacent applications including medical imaging diagnostics and movement/biomechanical assessment.

Strong in full SDLC ownership—from HPC/SLURM training and multi-GPU optimization to real-time user-facing deployment, monitoring, and CI/CD. Also delivers GenAI solutions with LLMs, RAG, and agentic workflows using GPT-4/Gemini/LLaMA and LangChain/LangGraph, and bridges technical outcomes with product, QA, and leadership to drive reliable, measurable impact.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

Work Experience

Senior AI/ML Engineer at COMPETE

February 1, 2026 - Present

Built and deployed an end-to-end AI-powered headstand challenge app using YOLOv8 pose estimation, including the full video-to-result pipeline (key-point detection, feature vector generation, frame classification, temporal hold validation, and pass/fail delivery). Trained and optimized a YOLOv8m-pose + MLP classifier achieving 99.88% test accuracy and 0.9987 F1. Implemented geometric posture-quality analysis using hip-knee-ankle key-point angles to provide interpretable form grades, plus temporal smoothing to reduce flicker and detect continuous hold segments. Developed OpenCV runtime overlays (HUD, score bars, stage progression, annotated outputs) and a modular scoring engine computing time/form/bonus components aggregated into final scores; iterated rule logic to reach perfect validation metrics before scaling to video-based scoring.

Graduate Research Assistant at University of Central Florida

August 1, 2025 - January 31, 2026

Presented research at AAG 2026 and UCF Student-Scholar Symposium 2026 (Judges’ Choice Award). Fine-tuned YOLO11m on 500K+ GoPro 4K frames for urban pedestrian detection, achieving [email protected] = 82.9% and surpassing a prior benchmark by +6.2% across 286 georeferenced scenes. Manually annotated data, resolved a VGG Image Annotator coordinate mismatch, and optimized YOLO training to recover mAP from 0.004 to 0.829 with significantly faster deployment. Engineered A100 batch inference generating 47M+ detections as CSV trajectories and used ByteTrack for identity-consistent tracking under occlusion. Performed RANSAC-based homography with <10px reprojection error and produced density heatmaps and georeferenced overlays, extracting behavioral metrics such as speed, crowd density, dwell time, and traffic compliance for public health mobility research.

Machine Learning Engineer at N2 Services Inc

May 1, 2025 - August 31, 2025

Deployed an end-to-end Yoga Pose Analysis System on GCP using MediaPipe and an MLP classifier, achieving 99.77% accuracy across 6 pose classes at 30 FPS with real-time inference. Built ML pipelines in TensorFlow/Keras/NumPy/Pandas/Scikit-learn for preprocessing, feature engineering, training, and evaluation. Extracted 66 hip-normalized features per frame from 88 videos with person-disjoint splits and applied augmentation to scale training data 9×. Implemented a biomechanical scoring engine based on anatomical rules using joint-angle accuracy and cosine similarity, plus bilateral pose detection and real-time overlays with smoothing/majority-vote logic. Created a reusable modular 5-stage pipeline with automated evaluation hooks, improved reproducibility, and delivered a GCP-deployed real-time inference pipeline supporting session analytics.

Technology Analyst / AI Engineer (Java & Python) at Infosys Limited

March 1, 2022 - July 31, 2024

Architected and deployed a production GenAI knowledge retrieval system using LLMs, RAG, and FAISS semantic search with sentence transformers, achieving 90%+ query accuracy under 2 seconds. Built and deployed an AI email automation agent using Google Gemini AI and Gmail API with RAG, achieving zero SLA breaches across 500+ daily requests. Developed production-grade RAG pipelines with LangChain/LangGraph (tool calling and state management) and deployed LLM applications on Vertex AI. Led server migration to AWS with CI/CD redesign and automated monitoring, reducing production incidents by 60% and sustaining 99% uptime. Translated ML outputs into business-readable reports, supported QA/product alignment, led code reviews, and mentored engineers. Delivered backend services via Spring/Flask/FastAPI and REST APIs as part of enterprise IT service and incident/problem/change management.

Application Developer at N2 SOFTWARE SERVICES PRIVATE LTD

August 1, 2014 - February 28, 2022

Designed and launched scalable Java and Python applications across full SDLC using Core Java, Spring Boot, and REST APIs/microservices (including Flask), improving delivery speed and latency metrics. Built reusable enterprise components with design patterns and implemented Java multithreading (Executor Service) to reduce latency in high-load scenarios. Integrated TensorFlow/Keras ML models with Python microservices and REST APIs for scalable inference workflows. Developed SOAP and REST web services using JAX-WS/JAX-RS, implemented CI/CD with Jenkins/Git/Maven/Gradle, and improved release cycle time. Designed Oracle and MySQL schemas with PL/SQL stored procedures and ETL migrations, reducing retrieval overhead and ensuring zero data loss. Led Agile planning, code reviews, and testing (JUnit), supporting multiple product launches and mentoring junior developers.