cv

Basics

Name Rohith Perumandla
Email rohith.perumandla.12@gmail.com
Phone +17739432531
LinkedIn www.linkedin.com/in/perumandla-rohith/
GitHub www.github.com/Rohith1-p
Summary AI/ML Engineer with 4+ years of experience building production-grade AI systems — multi-agent architectures, LLM applications, RAG pipelines, and full-stack SaaS platforms. Shipped healthcare AI SaaS end-to-end solo in under 4 months. Published researcher in conversational AI for dementia detection. Experienced in deploying scalable systems with guardrails, evaluation pipelines, observability, and cost optimization. Previously Data Scientist at SetuServ building NLP pipelines at scale.

Experience

  • Sep 2025 - Apr 2026
    Founding AI Engineer | R&D Venture Studio (Vector Referral)
    • Built a healthcare AI SaaS platform end-to-end solo in under 4 months — RAG pipelines, LLM agents, full-stack deployment, and production infrastructure.
    • Designed multi-agent AI architecture with guardrails, evaluation pipelines, and observability for production reliability.
    • Owned full lifecycle: model training, backend API, frontend UI, deployment, and iterative testing with industry advisors.
  • Sep 2023 - Dec 2025
    DePaul University, Graduate Research Assistant | AI Engineer
    • Designed and fine-tuned LLM-based conversational agents (Llama3, Phi-3) using PyTorch to assess cognitive decline in dementia patients, contributing to healthcare advancements.
    • Implemented machine learning algorithms for biomarker extraction from multimodal inputs (speech, text, audio) to enhance diagnostic accuracy.
    • Developed and deployed a scalable system on Google Cloud Platform (GCP) using Nginx, Docker, and WebSockets, enabling real-world testing, serving crucial Medicine applications.
    • Optimized system architecture, reducing response time by 75%, significantly enhancing real-time patient engagement and interaction.
    • Reduced GPU computational costs by 60% through model optimization, efficient inference strategies, and resource allocation improvements.
    • Engineered linguistic biomarker extraction modules, targeting six key impairments in dementia patients.
  • Jan 2022 - Jul 2023
    SetuServ, Data Scientist – Customer Management
    • Built and deployed an end‑to‑end scalable data science and ML system for extracting product reviews, ratings, and product details from e‑commerce websites.
    • Designed and automated MLOps data pipelines, reducing data processing time by 50% and optimizing relational database storage.
    • Performed advanced data preprocessing (cleaning, stopword removal, symbol elimination), EDA, and feature engineering, leading to a 15% improvement in sentiment classification accuracy.
    • Accelerated pilot project delivery by 70% using NLP and deep‑learning models (BERT, NER, zero‑shot classification, Word2Vec) from Hugging Face and spaCy.
    • Built and managed MLOps infrastructure, including a core ML server REST API in Python/Django, processing over half a million data points per day for enterprise‑scale solutions.
    • Improved data pipeline and quality checks by integrating Google Sheets with backend servers using Python, Google Apps Script, Django, scikit‑learn, and PyTorch.

Skills

  • AI & ML: LLM Fine-tuning, RAG, Multi-Agent Systems, Guardrails, Evaluation Pipelines, Prompt Engineering, LangChain, Hugging Face, LlamaIndex, Agentic AI
  • ML Frameworks: PyTorch, TensorFlow, Keras, scikit-learn, Transformers, NumPy, Pandas
  • Backend & Full-Stack: Python, Django, FastAPI, REST APIs, WebSockets, JavaScript, HTML/CSS, Swift (iOS/macOS), SwiftUI, SwiftData
  • Infrastructure & DevOps: AWS, GCP, Docker, Nginx, Kafka, Databricks, Git, CI/CD, Observability
  • Database: PostgreSQL, SQLite, MongoDB, Supabase, Vector Databases (Pinecone, Weaviate)
  • Analytics: Tableau, Excel, Data Visualization, Statistical Analysis

Licenses & Certifications

Projects

  • Healthcare AI SaaS Platform (Vector Referral)
    Built end-to-end solo in under 4 months. RAG pipelines, LLM agents, multi-agent architecture, full-stack deployment with guardrails and observability.
  • ProductivityAI (plantodo.co)
    macOS/iOS app with AI-driven task management, habit tracking, analytics, and cross-platform sync. Built with SwiftUI, SwiftData, and Supabase.
  • AftercareOS (aftercareos.com)
    Multi-layered AI automation platform for medical aesthetics. SMS-first patient aftercare, AI scheduling, and clinic workflow automation.
  • Brain Tumor Classification and Segmentation using Deep Learning
    Implemented deep learning models utilizing ResNet50 and convolutional neural network (CNN) architectures to accurately classify and segment brain tumors from medical images, achieving a segmentation score of 0.87