Aditya N.
Machine Learning Engineer
Experience
Jun 2023 - Present
2 years 4 monthsMachine Learning Engineer
jeevahealth.ai
- Designed, built and deployed end-to-end ML pipelines across mobile and web from prototyping to production, including real-time inference and model monitoring, ensuring CI/CD and observability.
- Built and scaled a chatbot with persistent memory (short and long term), rapid context switching, and personalized interventions mapped from a knowledge base using RAG techniques, serving 10K+ users with p95 latency of <500ms.
- Automated therapy session workflows by generating SOAP notes, treatment plans, and session summaries from live therapy sessions, using therapists’ feedback to improve models and context, and built an analytics dashboard to track user progress and clinician performance.
- Developed a voice/video enabled simulated learning platform simulating patients and sales/investment scenarios offering real-time feedback, scenario grading, and coach-led improvement tips; fine tuned LLama3-70B using LoRA, with ~1000 students using the platform and p95 latency of <100ms in text and <500ms for video/audio.
- Enabled chatbot driven goal tracking, auto-generated daily task plans, SAM scoring, and journal entries tailored to user behavior.
- Technologies: vLLM, Docker, MLflow, AWS SageMaker, RAG, LangChain, LangGraph, LlamaIndex, Unsloth, QLoRA, FAISS, ChromaDB, Elasticsearch, Hugging Face, OpenAI, Gemini, Llama, BERT, Transformers, scikit-learn, PyTorch, TensorFlow, Keras, PySpark, Pandas, SQL, R, Excel, Tableau, Power BI, XGBoost, Linux, Git, BitBucket, Jira, Node.js, Pinecone, ZepMemory, PostgreSQL, AWS EC2, 11Labs, Zapier
Jul 2021 - May 2023
1 year 11 monthsData Scientist
Genpact
- Collaborated in the MHRA COVID vaccine monitoring PVAI solution used to track vaccine adverse reactions.
- Built prediction modules for pharma clients using classification and NER, improving F1 score and accuracy by 75% across all source types.
- Built an encoder for different source types using AWS Textract, resulting in breadth-wise score improvements for all prediction fields.
- Built a multi-entity NER model extracting three primary fields using SpanBERT from Hugging Face, achieving an F1 score of 0.82 and accuracy of 0.95.
- Technologies: Python, TensorFlow, PyTorch, MLflow, OpenCV, Hugging Face, Docker, AWS SageMaker, AWS Textract, BERT, BitBucket, Jira, Linux
Jan 2021 - Jun 2021
6 monthsData Analyst
Foundation for Innovative New Diagnostics (FIND)
- Built and analyzed country models for diagnostic network optimization, including scenario modelling and cost-benefit analyses using LlamaSoft supply chain software.
- Assessed data, identified gaps, and cleaned and integrated data using R, Python, and SQL.
- Identified, analyzed, and interpreted trends from data and shared reports using Tableau.
- Received a pre-placement offer from the organization.
- Technologies: LlamaSoft supply chain software, MS Excel, SQL, R, Python, Tableau
Languages
English
AdvancedEducation
Oct 2017 - Jun 2021
Delhi Technological University
Bachelors in Technology · New Delhi, India · 8.2 CGPA
Delhi International Public School
Senior Secondary XII · New Delhi, India · 94.25%
Delhi International Public School
Secondary X · New Delhi, India · 10 CGPA
Certifications & licenses
Deep Learning Specialization
Machine Learning
Python Data Structures
Need a freelancer? Find your match in seconds.
Try FRATCH GPT More actions
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.