NIKITA DHARMADHIKARI

Applied AI Engineer

Baltimore, US.

About

Highly accomplished Applied AI Engineer with a Master's in Data Science, specializing in developing and deploying advanced AI/ML solutions. Proven expertise in building autonomous AI agent workflows, predictive forecasting models, and robust data pipelines, consistently delivering measurable results such as a 40% reduction in manual recruiter intervention and 20% overspend reduction. Seeking to leverage deep technical skills in LLMs, vector embeddings, and cloud infrastructure to drive innovation in a forward-thinking organization.

Work

HirexIT.ai

|

Applied AI Engineer

Remote

→

Summary

Spearheaded the engineering of advanced AI solutions, including a hybrid candidate-matching system and autonomous agent workflows, to optimize recruitment processes and enhance decision-making.

Highlights

Engineered a hybrid candidate-matching system utilizing Vector Embeddings and Vector Search (GCP/MongoDB Atlas), enabling deep semantic insight analysis for top-tier candidates.

Built autonomous AI agent workflows for interview automation with LLMs, function calling, and structured outputs (Pydantic), reducing manual recruiter intervention by 40%.

Designed robust prompt engineering pipelines with guardrails and fallback handling, ensuring reliable, production-grade AI decision-making.

Architected microservices on GCP Cloud Run to handle high-concurrency AI agent workflows, leveraging Cloud Trace for performance profiling and optimization.

Aptech

|

Data Scientist

Remote

→

Summary

Developed and deployed advanced ARIMA and Prophet forecasting models and conducted A/B tests to drive significant engagement improvements and operational efficiency.

Highlights

Built ARIMA and Prophet forecasting models to predict demand patterns, successfully reducing overspend by 20%.

Designed and evaluated A/B tests for feature launches, identifying statistically significant engagement improvements.

The RealReal

|

Cyber Security Analyst

SF, CA, US

→

Summary

Consolidated security logs into modeled warehouse views for near real-time anomaly monitoring; reduced false-positive alerts by 30% through rule tuning and API-based telemetry integration

Maryland Department of Health

|

Data Analyst

Baltimore, MD, US

→

Summary

Trained and deployed time series forecasting models in Azure ML Studio, integrating outputs into Power BI dashboards to inform strategic staffing and operational planning.

Highlights

Trained time series forecasting models in Azure ML Studio, informing staffing schedules and operational planning with accurate forecast outputs.

Deployed robust forecast pipelines and connected outputs to Power BI dashboards while maintaining critical data contracts.

Education

University of Maryland, Baltimore County

Baltimore, MD, United States of America
→
Master of Professional Studies Data Science

Cummins College of Engineering for Women

Pune, Maharashtra, India
→
Bachelor of Technology Computer Engineering

Publications

Enhanced Travel Experience using AI

Procedia Computer Science

Achieved 86% precision score in travel planning accuracy through prompt engineering, recommendation engines, and web scraping, published in Procedia Computer Science: ScienceDirect.

Skills

Programming Languages & Tools

Python (pandas, NumPy, scikit-learn, Pydantic), SQL, Git.

AI/Agentic/LLM

Autonomous AI Agents, Agentic Workflows, Function Calling, Tool-use, Prompt Engineering, LLMs, RAG Pipelines, LangChain, LlamaIndex, Vector Databases, Embeddings, Semantic Search.

Machine Learning

Scikit-learn, NLP, Predictive Analytics, Time Series Forecasting.

Cloud/Infrastructure

GCP (Cloud Run, Cloud Trace), MongoDB Atlas, Serverless Deployment.

Data Engineering

ETL pipelines, REST APIs, JSON/CSV/Parquet.

Frontend

Streamlit, Next.js.

DevOps

CI/CD, GitHub Actions.

Projects

AI Executive Agent

Summary

Built an autonomous agent that plans tasks and executes with a toolset via Streamlit interface, demonstrating agentic workflow design and function calling patterns, including an eval suite to measure task success and failure modes across labeled test sets.

Few-Shot Schema Transfer

Summary

Standardizes incoming CSVs to a target schema using few-shot prompting, Pydantic validation, and rule-based fallbacks, emitting typed JSON for downstream loads.

RAG-Powered Healthcare Automation

Summary

Developed a RAG pipeline to ground LLM outputs in clinical datasets, improving retrieval accuracy for decision-support automation.