I'm Srivathsav, a Software Development and ML engineer.
Here's a quick intro about me and what I love to do
My journey into AI/ML
My fascination with artificial intelligence began during my undergraduate studies in Computer Science at Koneru Lakshmaiah University. What started as curiosity about machine learning algorithms quickly evolved into a passion for building intelligent systems that can solve real-world problems. I dove deep into Python, data science libraries, and neural networks, spending countless hours experimenting with different models and architectures.
From Theory to Production
My first major breakthrough came at Samsung, where I worked as an ML Engineer focusing on Computer Vision. I optimized ResNet models across different frameworks, reducing inference latency by 40% while maintaining accuracy. This experience taught me the critical importance of model optimization and deployment in real-world scenarios, laying the foundation for my expertise in MLOps and production ML systems.
Expanding Horizons
Pursuing my Master's in Computer Science at New Jersey Institute of Technology has been transformative. It's where I deepened my understanding of advanced AI/ML concepts and connected theory with practical applications. The rigorous academic environment, combined with hands-on projects, helped me develop a more systematic approach to solving complex data science problems.
Current Focus
Currently, I'm working as a Data Scientist at WatchRx, where I build speech-to-text pipelines processing thousands of healthcare calls monthly. I'm also the Technical Founder of Moxium, an AI-powered educational platform that uses adaptive learning engines to personalize student experiences.
When I'm not immersed in data and models, you'll find me exploring the latest research papers, contributing to open-source ML projects, or experimenting with new AI architectures. I'm always excited to discuss the future of artificial intelligence and its potential to transform industries.
My work history and achievements timeline.
WatchRx
Mar 2025 - Present
Data Scientist
Built speech-to-text pipeline processing 1000+ hours of healthcare calls monthly using VAD, speaker diarization, and Whisper STT, achieving 95% accuracy for timestamped transcripts.
Shipped NLP summarization systems, reducing manual documentation by 80% and automating review for 500+ daily calls.
Implemented HIPAA-compliant data framework with role-based access controls and audit logging.
Improved transcription quality by 25% through confidence thresholds, language auto-detection, and intelligent re-decode queuing.
Established MLOps with MLflow tracking, drift monitoring, and CI/CD pipelines, reducing deployment time by 60%.
Verizon
Sep 2024 - Dec 2024
Data Scientist (AIOps / MLOps)
Built anomaly detection on 5TB+ Kubernetes logs daily using Isolation Forest and custom machine learning approaches, achieved 70% accuracy, reduced detection time by 45%.
Created self-healing recommender using RAG over 1000+ runbooks, reduced manual intervention 80%, cut resolution time from 2 hours to 15 minutes.
Orchestrated ETL and ML pipelines in Airflow/Spark, processing 10M+ daily events with streaming and batch scoring.
Deployed Kafka ingestion with schema registry, improved throughput 300% and achieved <100ms end-to-end latency.
Implemented Grafana dashboards tracking detailed model metrics with automated SLO alerting.
Samsung
Jun 2021 - Jun 2023
ML Engineer — Computer Vision
Optimized ResNet models across ONNX/TensorFlow/TFLite, cut inference latency 40% while maintaining 95% accuracy over 4000+ tests.
Implemented operator-level optimizations (fusion, graph optimization), achieving 30% memory reduction and 25% faster execution.
Built benchmarking suite profiling small models, guided quantization/pruning for 2x speedup on edge devices.
Automated CI/CD for model conversion with golden output validation, reduced regressions 90%.
Moxium
Mar 2025 - Present
Technical Founder
Built an adaptive learning engine with student/skill embeddings and DKT/Transformer to serve next-best lesson recommendations.
Designed a recommendation system for certification–competency fit using content tagging, SBERT embeddings, and learning-to-rank.
Implemented knowledge-gap detection with real-time features; auto-adjusted learning paths; tracked Top-k Recall, MAP, and time-to-mastery.
Set up comprehensive MLOps: MLflow experiments, data quality checks, drift monitoring, and offline/online evaluation.
Here's what sets me apart and makes me unique
Loading music...
Scrapbook
Currently Reading
Connections
Its Start of my Journey Lets See Who I Meet Along The Way
Stats coming soon
Subscribe to my newsletter
A periodic update about my life, recent blog posts, how-tos, and discoveries.
NO SPAM. I never send spam. You can unsubscribe at any time!
