01
95% retrieval accuracy
AskNEU — RAG Assistant
A question-answering assistant built over 500+ university web pages, so students can ask in plain language and trace an answer back to its source.
LangChainHugging FacePineconeAirflowGCP
Data Scientist · Software Engineer
Hi, I’m Yogita Bisht.
I build machine learning models and the data pipelines that power them; focusing on scalable cloud architecture, and how systems actually hold up under real-world usage.
01Transition story
From vehicle diagnostics to data science.
With a foundation in computer science and data engineering, my early work focused on building resilient backend architectures, scalable data pipelines, and high-reliability systems. At Mercedes-Benz R&D, I applied these core principles to engineer production data pipelines for high-volume vehicle data.
I expanded into data science during my Master’s at Northeastern, bridging software engineering with modern AI to build intelligent, end-to-end ML systems. During my time there, I worked as a research assistant training vision models for Skin Cancer detection alongside developing other ML systems.
500GB+ECU fault logs parsed
20%Fewer undetected diagnostic errors
95%Test coverage
25%Faster field issue resolution
02Selected work
A few projects I’ve worked on.
02
0.94 F1-score
Personalized Financial Recommendation System
A Streamlit app that matches user profiles against 26K+ financial products and generates personalized explanations through a RAG pipeline.
PythonXGBoostMLflowRAGStreamlit
03
54% → 94% JSON pass rate
Banking Audit Extraction via LoRA
Llama-3.2-3B fine-tuned with LoRA on 300 domain-specific audit records to extract structured JSON metadata from unstructured banking logs.
Llama 3.2LoRAFine-tuningJSONGGUF
04
95% sensitivity · 0.96 AUC-ROC
Melanoma Detection Research
Four hybrid CNN–Vision Transformer architectures benchmarked for melanoma detection on a severely imbalanced medical image dataset.
PyTorchViTCNNComputer Vision400K+ images
03Technical expertise
Tools I work with.
Languages & frameworks
PythonPandasNumPyScikit-learnSQLPyTorchFastAPIStreamlit
Machine learning & AI
LLMsFine-tuningRAGLangChainVector DBMLflow
Cloud, big data & DevOps
AWSGCPAirflowPySparkDockerCI/CDGitPower BIBash / Linux
04Credentials
Education
Northeastern University
Master of Science in Data Science
2024–2026 · GPA 3.6/4.0
Visvesvaraya Technological University
Bachelor of Engineering in Computer Science
2017–2021 · CGPA 8.5/10