👋Hi, I'm Rohit
Data Scientist & AI Engineer
I spend my days turning messy, real-world data into AI systems people actually use — multi-agent pipelines for healthcare, fraud detection watching 5M+ transactions a day, and LLM apps that save teams real hours. Currently building with Python, GCP, Azure & GPT-4o.
About Me
Data Scientist with 4+ years building production ML systems across healthcare, e-commerce, non-profit, and enterprise domains. Specializing in multi-agent AI systems, LLM-powered applications, product analytics, real-time anomaly detection, and scalable data engineering pipelines using Python, PySpark, Azure, GCP, and AWS.
With a background spanning Engineering (B.Tech from IIIT Chennai) to Information Systems (MS from University of Maryland — Smith School of Business), I bring a rare blend of engineering rigor and business context to every data problem.
class DataScientist: def __init__(self): self.name = "Rohit Ananthan" self.focus = ["GenAI & RAG", "MLOps", "Graph ML", "Product Analytics"] self.stack = {"lang": "Python", "cloud": ["GCP", "AWS"], "llm": "GPT-4o"} self.education = ["MS InfoSys @ UMD", "B.Tech @ IIIT Chennai"] def mission(self) -> Impact: # raw data in → business impact out return bridge(technology, business_value) # converged ✓
Experience
AI Engineer
CurrentVdart Inc.
Contract · Remote
- ▸Building multi-agent AI systems that automate document-heavy enterprise workflows for a healthcare client (specifics under NDA)
- ▸Working across retrieval-grounded evaluation, agent orchestration, and full-stack delivery — with AI coding agents as part of the development workflow
Data Scientist Consultant
Invision Global Tech Inc
Full-time · United States
Data Scientist
Community Dreams Foundation
Full-time · Remote
- ▸Built an AI-powered Legal & Compliance Assistant using GPT-4o with a RAG pipeline on LangChain, Pinecone, and ANN search — cutting manual review time by 40%
- ▸Deployed end-to-end ML pipelines on GCP Vertex AI and Dataproc with automated hyperparameter tuning via MLflow — improving demand forecasting accuracy by 15%
- ▸Implemented MLOps workflows using Vertex AI, Cloud Build, and GitHub Actions for automated model versioning, drift detection, and CI/CD across environments
- ▸Built a real-time fraud and anomaly detection system using Pub/Sub, Dataflow (Apache Beam), and XGBoost — reducing undetected fraud by 20%
- ▸Engineered scalable data pipelines with Dataflow, Composer (Airflow), and BigQuery — cutting ETL latency by 60%
- ▸Developed LLM-based apps for document summarization and Q&A using OpenAI APIs and Vertex Matching Engine for semantic search
- ▸Created predictive donor churn models with TensorFlow and Scikit-learn, deployed on Vertex AI with Looker Studio dashboards
Financial Analyst
The Premiere Group
Full-time · Columbia, MO · On-site
Technical Consultant – Course Renewal Automation
University of Maryland – Extended Studies
Internship · College Park, MD · Remote
Graduate Assistant
University of Maryland
Part-time · College Park, MD · On-site
Data Scientist
Kameleon Technologies
Full-time · Chennai, India
- ▸Engineered a real-time fraud detection pipeline using Neo4j graph database and XGBoost — processing 5M+ daily transactions with an 18% reduction in false positives
- ▸Built scalable ETL/ELT pipelines on PySpark and AWS (S3, Glue, EMR) to process terabytes of financial data — reducing pipeline processing time by 35%
- ▸Developed customer segmentation and churn prediction models using ensemble methods — driving a 12–15% improvement in customer retention across key segments
- ▸Established MLOps practices with MLflow experiment tracking, SageMaker model registry, and automated retraining workflows for production model governance
- ▸Delivered executive-facing Tableau dashboards for transaction monitoring, KPI tracking, and fraud trend analysis — adopted across operations and risk teams
Skills
Languages & Libraries
ML & AI
Data Engineering
Cloud & Infra
BI & Visualization
Databases & Search
Projects
Maez — A Digital Companion
A new kind of AI companion that grows, learns, and bonds with a single user for life. Runs locally on consumer hardware with persistent memory, a self-evolving cognitive loop, and proposal-based autonomy — the user owns the credentials, Maez owns the intent.
TrendScope
AI agent for content strategy — analyzes live YouTube trends, scores each by velocity, engagement, and competition, then delivers a ranked action plan with titles, hooks, and timing rationale via a ReAct reasoning loop.
Gym Aesthetic Trap
NLP research project using LDA topic modeling to analyze online discourse around SARMs and steroid usage — uncovering themes, risk perception patterns, and community sentiment from bodybuilding forums.
Thermal Error ML Modeling
Published Springer research on machine learning compensation strategies for thermal deformation in precision machine tools — achieving state-of-the-art accuracy in error prediction.
Education
Master of Science – Information Systems
University of Maryland
Robert H. Smith School of Business
📍 College Park, MD
Bachelor of Technology – Engineering
IIIT Chennai
Indian Institute of Information Technology
📍 Chennai, India
Certifications
Neo4j Graph Data Science Certification
Neo4j
AWS Certified AI Practitioner (AIF-C01)
Amazon Web Services
BCG Data Science Job Simulation
Boston Consulting Group × Forage
Publications
Mathematical Modeling of Thermal Error Using Machine Learning
Springer · Oct 6, 2022Research on thermal error modeling in machine tools using machine learning algorithms to identify the most effective compensation strategies for linear expansion and deformation caused by heat inputs from internal and external sources.
Model Card
// specifications
- parameters
- 4+ years of production experience
- architecture
- human × (ML + GenAI + product sense)
- context_window
- always open
- training_data
- healthcare · e-commerce · non-profit · enterprise
- fine_tuned_on
- GCP Vertex AI · AWS SageMaker
- inference_hardware
- coffee ☕ + RTX 4090
- temperature
- 0.7 — creative but reliable
- license
- open_to_hire
intended use
- ✓Shipping LLM/RAG systems to production
- ✓Real-time anomaly & fraud detection
- ✓Graph analytics on connected data
- ✓MLOps: versioning, drift detection, CI/CD
- ✓Turning ambiguous business questions into models
out of scope
- ✗Dashboards nobody looks at
- ✗Models that never leave the notebook
- ✗"We'll clean the data later"
- ✗Meetings that should have been a Slack message
// eval results (measured in production, not on a leaderboard)
known limitations
- · may overfit to interesting problems
- · inference quality degrades without coffee
- · cannot resist optimizing a slow query
Contact
Let's build something great together.
I'm currently open to Data Scientist, AI Engineer, and ML Engineer roles. Whether you're hiring, have a question about my work, or just want to talk data — I'd genuinely love to hear from you.