App LogoApp name
SW

Satish Wagh

Open to work3 years experience

AI/ML Engineer

Pune, IndiaTarget Roles: Backend Engineer • Full-Stack Developer • Frontend Specialist

AI/ML Engineer with 3.2 years of experience in GenAI and ML.

GitHub

Standing Rank

Rank Not Available

Developer Badges

No badges earned yet

Skills & Technologies

37 skills
pythonsqlbashlanggraphlangchainprompt engineeringraggpt-4azure openaiscikit-learnhugging face transformerssentence transformersnlpocrdeep learningxgboostpineconefaisschromadbelasticsearchfastapiflaskdjangorest apisgraphqlazureawsdockerkubernetesmlflowlangsmithstreamlitpostgresqlmongodbredispower bigit

Work Experience

AI/ML Engineer

Mobicool Software Solution Pvt. Ltd.

May 2023 - Present Pune, India
  • Developed and deployed end-to-end ML solutions (Credit Risk, OCR, Text Classification) that improved prediction accuracy and reduced manual work
  • Designed and deployed enterprise-scale Retrieval-Augmented Generation (RAG) systems using LangChain, LangGraph, Azure OpenAI, Pinecone, and FAISS, achieving ~95% answer relevance in internal evaluations.
  • Created efficient data pipelines with feature engineering, scaling, encoding, and outlier handling, speeding up model training and improving performance.
  • Tuned ML models (Logistic Regression, Random Forest, Gradient Boosting) using GridSearchCV and cross-validation to achieve high accuracy and stability.
  • Built multi-agent workflows using LangGraph and MCP for reasoning, planning, tool execution, and stateful orchestration across enterprise AI assistants.
  • Integrated and benchmarked GPT-4o, Claude, Gemini and Llama 3 across cost, latency, and response quality to support model selection for production workloads.
  • Implemented evaluation pipelines using RAGAS, LangSmith, and TruLens to improve retrieval quality, reduce hallucinations, and support continuous prompt optimization.
  • Deployed AI applications using Streamlit/FastAPI and integrated AWS/Azure services to deliver scalable, production-ready solutions.

Projects

Automated Resume Evaluation System

PythonLangChainLangGraphFAISSChromaDBSentence TransformersCross-EncodersOpenAI GPT-3.5/GPT-4oFastAPIStreamlit
  • Designed an end-to-end RAG pipeline document chunking, embedding generation, FAISS-based vector retrieval, cross-encoder re-ranking, and LLM synthesis to semantically match candidate resumes against job descriptions with high contextual accuracy.
  • Automated skill extraction, experience summarization, and candidate job fit scoring using GPT-3.5/GPT-4o with structured few-shot prompt engineering, reducing manual resume screening effort by 60%.
  • Implemented metadata filtering and hybrid keyword-semantic search to improve match relevance across resumes with varying formats, sections, and experience levels, increasing shortlist precision for recruiters.
  • Built an evaluation pipeline using RAGAS-style metrics to score matching accuracy and relevance, enabling continuous tuning of retrieval and ranking thresholds based on recruiter feedback.
  • Developed a FastAPI backend exposing resume parsing, embedding, and matching as async REST endpoints, integrated with a Streamlit frontend supporting real-time, multi-resume batch evaluation.
  • Containerized the application with Docker and tracked embedding model versions and evaluation experiments via MLflow, ensuring reproducibility and simplified deployment across environments.

Document Q&A Chatbot

Azure OpenAILangChainLangGraphFAISSChromaDBSentence TransformersCross-EncodersPythonFastAPIStreamlitLangSmith
  • Developed a document-grounded chatbot using Azure OpenAI and a Retrieval-Augmented Generation (RAG) architecture, enabling accurate contextual reasoning and multi-turn Q&A across heterogeneous enterprise documents.
  • Built an automated ingestion and embedding pipeline converting PDF, DOCX, Excel, and TXT files into vector representations, with metadata tagging for source tracking and improved retrieval filtering.
  • Implemented hybrid retrieval combining FAISS vector search with Sentence Transformers and cross-encoder re-ranking, reducing irrelevant context and improving answer precision across multi-document corpora.
  • Added Redis-based caching for frequently asked queries and repeated document lookups, lowering average response latency for high-traffic query patterns.
  • Integrated LangSmith for tracing conversation flows and evaluating response groundedness, helping identify and reduce hallucinated or unsupported answers during testing.
  • Built a Streamlit interface supporting chat history, multi-document upload, and source citation display, and containerized the full application with Docker for consistent deployment across environments.

Credit Risk Analytics Platform

PythonPandasNumPyScikit-learnXGBoostGridSearchCVMLflowStreamlit
  • Built an end-to-end credit risk prediction pipeline covering data cleaning, missing-value handling, categorical encoding, and feature engineering.
  • Trained and compared multiple classification models Logistic Regression, Random Forest, Gradient Boosting, and XGBoost to identify the best-performing algorithm for default risk prediction.
  • Performed hyperparameter tuning using GridSearchCV with stratified cross-validation, achieving the best ROC-AUC score of 0.79 using a tuned Gradient Boosting model.
  • Evaluated models using Accuracy, ROC-AUC, Precision-Recall, Confusion Matrix, and Classification Report to select the optimal model balancing false positives and false negatives for lending decisions.
  • Engineered derived features such as debt-to-income ratio and credit utilization trends, improving model discriminative power and reducing misclassification of high-risk applicants.
  • Tracked experiments, model versions, and evaluation metrics using MLflow, and built a Streamlit dashboard for visualizing model performance and risk scores for business stakeholders.

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

AI/ML Engineer with 3.2 years of experience designing, building, and deploying production-grade GenAI applications. Experienced in developing RAG systems, LLM-powered assistants, and multi-agent workflows using LangChain, LangGraph, Azure OpenAI, and FastAPI.

Bachelor of Science in Statistics

K.T.H.M. College, Nashik (2020 - 2023)

Master of Science in Statistics

K.T.H.M. College, Nashik (2023 - 2025)

Languages: English, Hindi
Satish Wagh - Profile | Swiftcruit