Harsh Joshi
Open to work3 years experienceData Scientist
Data Scientist | Machine Learning | Artificial Intelligence
Standing Rank
Rank Not Available
Developer Badges
No badges earned yet
Skills & Technologies
43 skillsWork Experience
Master Thesis Intern
Fresenius Medical Care
- Developed defect detection models for manufacturing quality control by training CNN and Transformer architectures on production images, increasing detection accuracy by 12%.
- Improved model robustness for fiber inspection by applying image preprocessing and augmentation techniques, increasing performance consistency by 21%.
- Deployed computer vision models to cloud environments by packaging inference pipelines on Azure, reducing deployment turnaround time by ~30%.
Working Student Data Science/Machine Learning
Fresenius Medical Care
- Built predictive models for dialysis patient outcomes using 50,000+ clinical records by engineering features, applying dimensionality reduction, and training models in Python on Azure with MLflow, improving prediction accuracy by 12%.
- Developed production-ready data pipelines for structured clinical records by standardizing preprocessing, validation, and evaluation workflows, reducing model error by 25%.
- Delivered decision-support dashboards for clinicians by integrating model outputs into Power BI and visualizing trends across patient records, accelerating clinical insights by 33%.
Machine Learning Engineer
Teksun Inc
- Designed real-time human activity recognition systems for elderly care by processing wearable sensor data and training TensorFlow Lite models deployed on ESP32 and Android devices, achieving 85% classification accuracy.
- Developed ADAS helmet system using YOLO and custom CNNs for pedestrian, vehicle, and traffic sign detection; deployed on Jetson Nano for edge inference, reducing false prediction latency by 12%.
- Implemented industrial safety monitoring systems by training computer vision models for fire, smoke, and PPE detection and deploying real-time alerting pipelines across sites, with hazard detection accuracy of 78%.
- Integrated domain-adapted LLM capabilities into production applications by fine-tuning LLaMA model and collaborating with product teams on deployment.
Projects
Research Paper Q&A Assistant
- Architected and implemented a RAG-based Q&A assistant over 25+ research papers by structuring text chunking, SentenceTransformer embeddings, LangChain orchestration, and ChromaDB indexing, improving answer relevance by 25%.
- Integrated and deployed Perplexity-based LLMs into a Streamlit application for interactive querying, increasing successful query resolution by 20%.
Weather Bot
- Engineered and deployed a conversational AI chatbot for real-time weather forecasts by integrating external APIs and automating Python pipelines for data parsing and response generation, reducing response latency by 40%.
- Integrated and operationalized intent handling with Google Dialogflow and deployed via Git and Ngrok, increasing successful user intent resolution by 30%.
Application Assistant (Agentic AI Job Research Tool)
- Architected and deployed an end-to-end agentic AI pipeline using LangGraph to orchestrate a multi-step agent graph that autonomously scrapes job postings from any URL, extracts company metadata, performs web research via Tavily search, parses CV documents, and generates structured match reports covering skill gaps, highlight recommendations, and cover letter strategy.
- Integrated Groq-hosted LLaMA3-70b for low-latency LLM inference, implemented persistent long-term memory with SQLite to cache job research and eliminate redundant API calls across sessions, and deployed an interactive Streamlit interface for real-time job-to-CV matching with actionable output.
Leaderboard Standings
Leaderboard Position Pending
Global test scores, peer standing percentiles, and algorithm leaderboard ranks are updated dynamically.
Assessment Highlights
Assessments Not Completed
Coding evaluations, system assessment results, and conceptual score badges will appear here after taking a test.
AI Collaboration Score
AI Collaboration Score Pending
Developer coding behavior, assistant cooperation, and AI pair-programming indicators are evaluated during live coding sessions.
Role Compatibility Profile
Role Compatibility Analysis Pending
Custom matching reports, candidate role compatibility percentiles, and core engineer strength profiles are processed once conceptual code screenings are complete.
Achievements
Achievements Not Earned
Special rewards, developer badges, system recognition certificates, and conceptual milestones will display here.
About Details
Professional Bio
Data Scientist with 3 years of experience building, optimizing, and deploying deep learning and ML systems across healthcare, industrial, and automotive domains. Strong background in computer vision, LLM fine-tuning, RAG, and edge AI. Proven ability to deliver measurable business impact.
M.Sc. Artificial intelligence
IU International University of Applied Sciences (2023 - 2026)
B.Eng. Information technology
Birla Vishvakarma Mahavidyalaya (2017 - 2021)