Hi, I'm Rushendra!!

I build applied AI systems that turn messy, unstructured information into dependable products and workflows.

My strongest work sits at the intersection of LLM applications, agentic workflows, retrieval, evaluation, and backend systems. I care about the part after the demo: incomplete context, noisy data, latency constraints, and domain-specific failure modes.

Much of my experience has been in healthcare and biomedical AI, including document intelligence, medical imaging, and clinical workflow automation. That domain shaped how I engineer AI systems: clear abstractions, careful evaluation, reliable surrounding infrastructure, and enough domain understanding to be useful in practice.

Alongside production engineering, I bring 3+ years of research experience in computer vision, robustness, interpretability, and trustworthy ML. I like building AI systems that are ambitious, practical, and disciplined enough to survive real-world use.

Open to ML, applied AI, and backend roles.
Portrait of Rushendra Sidibomma
Resume CV LinkedIn Scholar GitHub
Education

MS Computer Science

University of Minnesota Twin Cities

2025 - Present (Expected May 2027)

BS Computer Science with Honors

IIIT Sri City

2020 - 2024

Skills

Languages & AI Coding Tools

  • Python
  • Go
  • C/C++
  • SQL
  • JavaScript
  • Java
  • Bash
  • Codex
  • Claude Code

LLMs & Agents

  • Structured Extraction
  • Document AI
  • Vision-Language Models
  • RAG
  • LLM Evaluation
  • Agent Orchestration
  • Tool Use

Machine Learning

  • PyTorch
  • Transformers
  • Scikit-Learn
  • MLOps
  • Computer Vision
  • RLHF/RLVR
  • Fine-tuning
  • LoRA
  • Numerical Methods

Backend & Systems

  • Django
  • FastAPI
  • Node.js
  • REST APIs
  • Concurrency
  • Distributed Systems
  • Microservices
  • Grafana
  • Prometheus
  • Telemetry

Databases & Infrastructure

  • PostgreSQL
  • MySQL
  • MongoDB
  • Redis
  • Elasticsearch
  • AWS
  • Azure
  • Docker
  • CI/CD
  • Linux
  • Git
  • HPC

Experience

Present May 2026

AI Development Intern

BioStack Platforms (YC P26)

May 2026 - Present

San Francisco, CA

Designing reinforcement-learning environments and evaluation scenarios for LLM agents in clinical workflows.

  • LLM Agents
  • RLHF/RLVR
  • Clinical AI
  • Evaluation
View details
  • Designed and implemented 15+ reinforcement-learning environments for LLM agents, converting longitudinal clinical workflows into measurable states, actions, rewards, and task success criteria for post-training.
  • Generated 300+ LLM-agent scenarios across reasoning, retrieval, and tool-use workflows, using structured traces and scoring rubrics to support RLHF/RLVR data generation and failure analysis.
  • Improved LLM-agent task success rate by 18% and reduced unsupported responses by 25% by converting failure traces into targeted red-teaming cases, reward refinements, and post-training data signals.
May 2026 January 2026

Graduate Research Assistant

Masonic Cancer Research Center, University of Minnesota

January 2026 - May 2026

Minneapolis, MN

Built microscopy-image processing and analysis pipelines for multiplex histology datasets.

  • Microscopy
  • PyTorch
  • RAPIDS
  • Segmentation
  • Cancer Research
View details
  • Engineered Python pipelines for 230GB+ high-resolution microscopy images, handling preprocessing, segmentation, and structured feature extraction across diverse multiplex histology datasets.
  • Fine-tuned deep learning-based segmentation models using PyTorch to extract features from 1M+ cells; improved the segmentation accuracy of standard Cellpose models by 10% on internal lab-generated imaging datasets.
  • Developed GPU-accelerated clustering and analysis workflows with RAPIDS, Pandas, and NumPy, reducing manual review time by 60% and improving cell identification and classification reliability.
July 2025 July 2024

Machine Learning Engineer

AutomationEdge Technologies

July 2024 - July 2025

Pune, India

Built production document-intelligence and agentic extraction systems for clinical workflow automation.

  • Document AI
  • OCR
  • VLMs
  • Django
  • PostgreSQL
View details
  • Built a production document-intelligence platform combining OCR, VLM-based layout handling, agentic LLM extraction, and schema validation to convert unstructured clinical documents into validated records; reduced patient intake errors by 40%.
  • Fine-tuned a VLM-based layout classifier to detect multi-column layouts, checkboxes, and other complex page formats, routing only high-value pages to frontier vision models and reducing vision-model inference costs by over 50%.
  • Devised keyword-based page pruning to skip 60% of irrelevant pages before extraction, cutting model-processing costs by 40% while preserving field-level recall on gold-standard documents.
  • Implemented Django REST APIs and PostgreSQL-backed workflow state for upload, parsing, extraction, validation, audit logs, and RPA handoffs; kept large-document latency under 90 seconds across 20,000+ monthly requests.
  • Built a pytest-based regression suite over gold-standard medical documents to validate schema correctness and field-level extraction quality despite nondeterministic LLM outputs; kept extraction error rate under 10%.
May 2025 May 2023

Mitacs Fellow

Trustworthy AI Lab, Toronto Metropolitan University

May 2023 - May 2025

Toronto, Canada

Researched scalable neural-network robustness verification under a competitive Mitacs fellowship.

  • Trustworthy AI
  • Certified Robustness
  • Optimization
  • Mitacs
View details
  • Received a $10,000 fellowship with an acceptance rate below 8% globally among 25,000 applicants.
  • Built scalable verification frameworks to evaluate neural network reliability under adversarial conditions, reducing certification runtime by 70% compared to prior methods.
  • Designed optimization-based certification pipelines that improved computational efficiency by 47% while maintaining less than 2% deviation from comparable high-cost state-of-the-art methods.
  • Developed reproducible benchmarking methods to compare model robustness across architectures, supporting systematic model validation and safety assessment of deep learning models.
January 2024 September 2023

Backend Engineering Intern

Eyemote Vision

September 2023 - January 2024

Remote

Built backend ML inference services for diabetic-retinopathy screening from retinal fundus images.

  • Computer Vision
  • PyTorch
  • Healthcare
  • APIs
View details
  • Engineered a backend ML inference service in Python for detecting diabetic retinopathy from retinal fundus images, exposing deep learning model predictions via APIs for integration into clinical screening workflows.
  • Trained and deployed computer vision models in PyTorch to grade severity of diabetic retinopathy on large-scale retinal imaging datasets, optimizing inference for low-latency, near real-time predictions.
  • Integrated the ML system with partner eye clinics' internal systems to enable automated image ingestion and delivery of diagnostic predictions in production healthcare environments, shipping the system as the minimum viable product.

Projects

Distributed Rate Limiter Gateway

Production-style API gateway with Redis-backed distributed rate limiting, tenant policies, observability, and load-testing hooks.

Stack
  • Go
  • Redis
  • PostgreSQL
  • API Gateway

AfterVisit AI Voice Assistant

Voice-first post-visit care assistant that explains clinical records, answers grounded questions, and triggers consent-first follow-up actions.

Stack
  • Next.js
  • TypeScript
  • WebSockets
  • Grok Voice
  • Claude

Referral Intake & Schedule Automation

Agentic intake workflow for home-health referrals that turns messy discharge PDFs into validated, scheduling-ready referral records.

Stack
  • React
  • FastAPI
  • Pydantic
  • PDF/OCR
  • Browser Agents

CartScout

RL-oriented browser shopping environment that gathers purchase-critical context, cites evidence, and stops before checkout.

Stack
  • Python
  • Browser Agents
  • RL
  • HUD
  • Evaluation

Clinical RL Environments

Custom Gymnasium environments for clinical evidence-gathering agents that learn to inspect notes under a reading budget.

Stack
  • Python
  • Gymnasium
  • Clinical NLP
  • RL
  • Evaluation

Research Outputs

2026

IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)

Cascading Robustness Verification: Toward Efficient Model-Agnostic Certification

A model-agnostic robustness verification framework that progressively applies multiple verifiers to improve certified accuracy while reducing runtime.

Maleki, M., Sidibomma, R., Adibi, A., and Samavi, R.

2025

CHiPSAL Workshop, COLING

Hate Speech Detection and Target Identification in Devanagari Languages via Parameter Efficient Fine-Tuning of LLMs

A parameter-efficient LLM fine-tuning approach for low-resource hate speech detection and target identification in Hindi and Nepali.

Sidibomma, R., Patwa, P., Patwa, P., Chadha, A., Jain, V., and Das, A.

2023

IEEE International Symposium on Electronic Systems Devices and Computing

Learning Semantic Representations and Discriminative Features in Unsupervised Domain Adaptation

A domain adaptation method that aligns semantic representations while improving discriminative target-domain features.

Sidibomma, R. and Sanodiya, R. K.

Talks

Rushendra presenting an AI/ML talk for freshmen students.
August 2024

AI/ML Talk for Freshmen Students

IIIT Sri City · Andhra Pradesh, India

Introduced freshmen to AI/ML and later mentored 20 students beginning their machine learning journey.

Rushendra speaking at the Workshop on Advances of Deep Learning and Applications.
December 2023

Workshop on Advances of Deep Learning and Applications 3.0

IIIT Sri City · Chittoor, India

Helped host a four-day deep learning workshop featuring speakers from IIT Delhi, Nvidia, Samsung, Google, Adobe Research, and other organizations.

Rushendra with the Trustworthy AI Lab group after a research summary presentation.
July 2023

Research Summary Presentation

Trustworthy AI Lab, Toronto Metropolitan University · Toronto, Canada

Presented research progress from a 12-week assistantship on certified robustness for multilayer neural networks.