Looking for a job

Varun Ganjigunte Prakash

@varunganjiguntepraka

Message

I build production multimodal computer vision and AI systems for clinical video, speech, and document understanding.

India

Message

What I'm looking for

I'm looking to build production computer vision and applied AI systems, especially for multimodal video understanding, clinical healthcare, speech, and document intelligence.

I've built production AI systems at Frontera Health for clinical therapy video, combining multimodal video-audio inference, object detection, speech biometrics, and document AI.

I productionized scream/cry detection across 59.1 hours of therapy video and built a speaker-signature pipeline that achieved 92.2% mean person-level accuracy in clinic evaluations. I also fine-tuned RF-DETR for an ABLLS vision parser, reaching 0.936 AP50 on an unseen test split.

Previously, I led machine learning and computer-vision work at CogniAble, where I trained and deployed action-recognition models for children's behaviors and led nine production computer-vision models. At L&T Technology Services, I developed edge-inference applications, OCR automation, and OpenVINO-accelerated deep-learning workflows.

My work spans multimodal video understanding, clinical screening, treatment personalization, NLP, and computer vision research. I've authored five peer-reviewed publications on autism assessment, human action recognition, and robotics.

Experience

Work history, roles, and key accomplishments

Senior Research Scientist

Frontera Health

Jan 2025 - Sep 2026 (1 year 8 months)

Led evaluation of multimodal and pose-based approaches for rare maladaptive behaviors in long therapy videos, identifying a candidate cascade that removed 91.5% of normal chunks while retaining 73% of positives. Productionized multimodal scream/cry detection and speaker-signature pipelines, and fine-tuned RF-DETR for assessment grids.

VideoMamba YAMNet HumanOmni RGB+Pose SileroVAD Pyannote WeSpeaker RF DETR Gemini Vision PDF Rasterization Typed Schemas Regression Testing

Senior Computer Vision Engineer

Frontera Health

Dec 2023 - Dec 2024 (1 year)

Contributed to production social-engagement computer-vision stack including person tracking, child/adult detection, gaze models, and Gemini-based engagement summarization. Improved social-gaze estimation and clinical-video ASR.

YOLOv4 WHENet Gemini ASR Language Detection Gaze Estimation People Tracking

Senior Engineer, Data Science

Colt Technology Services

Mar 2023 - Dec 2023 (9 months)

Implemented and integrated NLP models for contract data mining, sentiment analysis, and opportunity prioritization. Evaluated scalable NLP techniques and ran biweekly ML/NLP sessions to drive adoption.

NLP Data Mining Sentiment Analysis Prioritization

Head Machine Learning and Technology Architect

CogniAble

Jul 2022 - Feb 2023 (7 months)

Researched and implemented computer vision approaches for clinical screening. Implemented a neural network-based recommender system for treatment personalization and skill enhancement for patients with neurodevelopmental disorders.

Computer Vision Recommendation Systems Neural Networks

Machine Learning Engineer - Computer Vision

CogniAble

Mar 2021 - Jul 2022 (1 year 4 months)

Trained and deployed a custom action-recognition network for children's behaviors and led nine production computer-vision models. Built object classifiers/trackers for clinical skill recognition and large-scale video training/inference pipelines.

Action Recognition Classification Trackings Video Processing Recommenders Graph Methods

Engineer

L&T Technology Services

Jun 2018 - Mar 2021 (2 years 9 months)

Developed three computer-vision edge-inference applications for a leading AI/semiconductor customer, covering detection, tracking, and on-device deployment. Designed an OCR pipeline for HMI screen automation and accelerated customer deep models with OpenVINO.

Computer Vision Edge inference Détection Trackings OCR OpenVINO Deep Learning