F5 Makes Hiring Easy
We handle the recruitment process, screening and shortlisting candidates.
Vishal Dev
Data Scientist
Data Scientist
Data scientist and AI-ML & NLP Engineer with over 7+ years of proven expertise in designing and deploying advanced models across industries such as FMCG, Wealth Management, Finance, and Entertainment. Adept at leveraging cutting-edge AI technologies, including LangChain, LLMs, and vector databases, to deliver impactful solutions that drive business growth and efficiency.
PREVIOUSLY AT
Siemens Pune
LOCATION
Lucknow, India
AVAILABILITY
Full-time
Skills
Agglomerative Clustering
AWS
BERT
Bertopic
BM25
CNNs
CoT
Data Mining
Decision Trees
Docker
Falcon
Gaussian Mixture Model)
Glove
GPT
GPT-3.5
+22 More
Work Experience
Data Scientist
Siemens Pune
September 2023 – Present
- Engineered a chatbot using LangChain RAG with a custom-built FAISS vector store for real-time IoT data insights.
- Created SQL-generating LangChain ReAct agents for querying building metrics and built custom Plotly chains for visualization.
- Designed a custom transformer architecture (16 attention heads, 3 encoder layers) for IoT device description generation.
- Implemented advanced clustering and translation workflows, including UMAP with silhouette scoring for Finnish language tickets.
TECHNOLOGIES
Python, Supervised (KNN, SVM, Decision Trees, Logistic Regression, XGBoost); Unsupervised (KMeans, Agglomerative Clustering, Gaussian Mixture Model), Spacy-NER, NLTK, BERT, Glove, BM25, RNNs, LSTMs, CNNs, Transformers, GPT, GPT-3.5, Llama, Falcon, OPT, MPT-7B, PyTorch, Scikit-learn, LangChain, AWS, Docker, Bertopic, LDA, NMF, MongoDB, Topic Modeling, Data Mining, LangChain (ReAct, CoT, RAG)
AI & ML Engineer
Persistent Systems Pune
August 2021 – September 2023
- Fine-tuned Falcon 7B and MPT-7B for extracting high-net-worth individual (HNI) information, improving decision-making efficiency.
- Automated balance sheet extraction for 100+ companies, reducing consultant workload by 40% and saving $50k annually.
- Implemented multi-label article classification using Sentence-BERT and XGBoost, achieving an F1 score of 91%.
- Developed a lead mapping feature using Neo4j, increasing lead conversion rate by 10% and generating $2.5M in revenue.
TECHNOLOGIES
Python, Supervised (KNN, SVM, Decision Trees, Logistic Regression, XGBoost); Unsupervised (KMeans, Agglomerative Clustering, Gaussian Mixture Model), Spacy-NER, NLTK, BERT, Glove, BM25, RNNs, LSTMs, CNNs
Data Science Consultant
Capgemini Pune
April 2021 – August 2021
- Delivered data science solutions, including fraud detection models and advanced resource allocation frameworks.
TECHNOLOGIES
Python, Spacy-NER, NLTK, BERT, Glove, BM25, RNNs, LSTMs, CNNs, Transformers, GPT, GPT-3.5
