I build AI that holds up in production.
PhD in Physics with over 8 years of industry experience in Generative AI, Agentic AI Workflows, Traditional ML/AI and MLOps. Proven track record of leading cross-functional teams, driving innovation, and delivering scalable AI solutions.
- 8+
- years building industry AI
- PhD
- physics · neural networks
- 4
- peer-reviewed papers
A physicist who ships.
I started in physics — a PhD at Ohio University, modelling neural networks and the dynamics of living systems. That work set the habit I still keep: respect the evidence, and build things that survive contact with the real world.
Today I lead generative and agentic AI at Discover — fine-tuning LLMs, designing multi-agent workflows, and standing up the MLOps that keeps them honest in production. I still teach and publish on the side, mostly computer vision for brain-organoid research at Ohio University.
I'm passionate about pushing the boundaries of what's possible with AI and machine learning. My journey spans academia and industry, where I've led teams in developing innovative solutions that have real-world impact.
- Based
- Lewis Center, Ohio, USA
- Doctorate
- Ph.D. Physics (Neural networks)
- Institution
- Ohio University, Athens, Ohio
- Focus
- Generative AI · agentic systems · MLOps
Where the work happened.
Manager/Lead - Generative AI
Discover
Apr. 2022 - Present
- Pioneered agentic AI workflows, automating customer experience and operational processes and generating comprehensive model documentation via LLM pipelines, resulting in a 75% reduction in manual effort
- Fine-tuned LLMs for company-specific datasets: applied LoRA, QLoRA, and quantization techniques to enhance local inference throughput by 40% and reduce GPU memory usage by 25%
- Implemented VLLM-based inference optimization, boosting custom LLM throughput by 50% and slashing latency for production endpoints
- Mentor and Lead a team of several direct and indirect reports, cultivating talent and ensuring delivery of high-impact AI solutions
- Designed and maintained robust MLOps pipelines with CI-CD, monitoring, and custom API endpoints
Senior Applied AI - ML Associate
JP Morgan Chase
Apr. 2017 - Apr. 2022
- Developed fraud detection AI/ML models, significantly enhancing the bank’s ability to identify and mitigate fraudulent activities.
- Engineered advanced feature extraction and seasonality methods, refining predictive capabilities and improving model reliability.
- Researched and implemented scalable neural network and NLP solutions (CNNs, LSTMs, BERT), advancing the processing of both structured and unstructured financial data.
- Compiled comprehensive technical documentation for end-to-end model development pipelines, ensuring transparency, audit readiness, and regulatory compliance.
- Collaborated with business and IT stakeholders to translate requirements into production-ready solutions, deploying and troubleshooting models for seamless integration.
Adjunct Professor
Ohio University
Nov. 2023 - Present
- Researched and taught advanced computer vision and segmentation methods for organoids, contributing to cutting-edge academic publications.
- Developed agentic AI research workflows, automating dataset curation and model retraining for organoid segmentation studies.
- Deployed reproducible research environments using Docker, AWS SageMaker, and ECR, enabling seamless collaboration and scalability.
- Developed a tri-agent fallback system, boosting YOLO-like detection recall and robustness in complex scenarios.
- Mentored PhD candidates, guiding their research on deep learning applications in biology.
Data Scientist
Nationwide Children's Hospital
Jun. 2016 - Apr. 2017
- Developed deep learning bioinformatics tools for real-time analysis of infant sensor data, enabling early detection of physiological anomalies.
- Designed and implemented waveform analysis algorithms to identify feeding disorders in neonatal patients using ML-driven pattern recognition.
- Applied fast Fourier transform and statistical methods to extract actionable clinical insights from biosensor signals.
- Collaborated with pediatric research teams to translate analytical findings into improved care protocols and treatment strategies.
- Documented analytical workflows and presented results in internal reviews, fostering adoption of AI-driven monitoring solutions.
Things I built outside the day job.
Small, sharp tools — most of them agentic, most of them open source.
Peer-reviewed work.
From cellular reprogramming to graphene chemistry to brain-organoid maturation.
- In review
Studying Brain Organoids Survival rates using automated segmentation methods
Pusuluri et.al. · 2025 (to be submitted)
- Published
Electrophysiological maturation of cerebral organoids correlates with dynamic morphological and cellular development
Pusuluri et.al. · Stem cell reports 15 (4), 855-868, 2020
- Published
Cellular reprogramming dynamics follow a simple one-dimensional reaction coordinate
Pusuluri et.al. · Physical Biology 10.1088/1478-3975/aa90e0, 2017
- Published
Role of deoxy group on the high concentration of graphene in surfactant/water media
Pusuluri et.al. · Royal Society of Chemistry (RSC Advances), 2012
What I reach for.
mlops
- Docker
- AWS SageMaker
- Amazon ECR
- CI-CD
- APIs
- VLLMs
- Python
- TensorFlow
- PyTorch
- Keras
- SQL
- Bash
agentic A I
- Agentic Tools
- MCP servers
- LangGraph
- CrewAI
- Multi-Agent Systems
leadership
- Team Management
- Mentoring
- Cross-functional Collaboration
- Stakeholder Engagement
generative A I
- Transformer models
- Gemini
- Claude
- OpenAI
- Llama
- Mistral
- RAG
- Chain-of-Thought
Let’s talk about the work.
Generative AI strategy, agentic workflows, production MLOps, or research — reach out.
- hello@saiteja.ai
- sai-teja-pusuluri
- GitHub
- sai19872000
- Based
- Lewis Center, Ohio, USA