Looking for a job

Kartik Bansiwal

@kartikbansiwal

Message

AI Engineer and Data Scientist building production GenAI, risk scoring pipelines, and applied machine learning systems.

India

Message

What I'm looking for

I’m looking for an Analyst / Data Scientist / ML Engineer role where I can build deterministic, production-grade GenAI and scalable ML pipelines—bringing measurable impact (latency, cost, adoption) and collaborating with teams to deploy models responsibly.

I’m an AI Engineer and Data Scientist focused on turning GenAI and machine learning into deterministic, production-ready decision systems. I currently work at L&T Finance, where I architect and deploy end-to-end analytics and risk workflows.

At L&T Finance, I built a Multi-Agent GenAI system using LangGraph and Vertex AI to orchestrate bureau and banking analysis with modular agents for retrieval, standardization, and policy reasoning. I also designed and deployed Bureau Analyzer and Banking Analyzer for SME and PL loans, delivering consolidated outputs like risk flags, deviations, and calculated variables—reducing TAT by 70% and driving 95% adoption across underwriting teams.

I develop scalable data pipelines, including a production-grade Vertex AI risk scoring pipeline processing 9.6 Cr+ monthly transactions for 2.1 Lakh customers with fully automated ingestion, feature engineering, batch scoring, and datamart publishing. I’ve improved efficiency with a 99.4% cost reduction and 91% runtime improvement (8.5 hrs → 45 mins) using auto-scalable batch orchestration.

Before this, I strengthened my research and modeling foundation through ML internships. At NTU Singapore, I optimized speaker diarization with Pyannote 3.1 and achieved a 24.3% DER; at NITI Aayog, I performed SQL/NumPy-Pandas policy analytics; and at Ignitus, I worked on CNN defect classification, PCA-enhanced regression, and LSTM Seq2Seq summarization with attention.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

L&T Finance

Jul 2025 - Present (11 months)

Architected a multi-agent GenAI system with LangGraph and Vertex AI to orchestrate deterministic bureau and banking analysis, cutting underwriting turnaround time by 70% and driving 95% adoption. Built and deployed production Vertex AI batch risk scoring and a real-time case allocation engine using Python and BigQuery for automated loan routing and faster credit operations.

LangGraph Vertex AI Multi Agent Systems Python Risk Scoring BigQuery Feature Engineering Orchestration

Machine Learning Research Intern

Speech Lab, NTU Singapore

Jul 2024 - Dec 2024 (5 months)

Built and optimized a multi-speaker speaker diarization pipeline using VAD, OSD, and speech separation, and benchmarked Pyannote 3.1 on DIHARD3. Achieved a 24.3% DER after hyperparameter tuning and developed end-to-end training, evaluation, and inference pipelines in Python.

Speaker Diarization VAD OSD Speech Separation Pyannote 3.1 DIHARD3 Hyperparameter Tuning Python Benchmarking ML Pipelines

Data Analyst Intern

NITI Aayog

May 2024 - Jul 2024 (2 months)

Performed comparative analysis of 140+ learning outcomes for aspirational vs non-aspirational districts using the NAS dataset. Ran saturation analysis across 49 indicators and correlated performing states using SQL and NumPy/Pandas for data-driven policy analytics.

SQL NumPy Pandas Data Analysis Saturation Analysis

Machine Learning Engineer Intern

Ignitus

Jun 2023 - Sep 2023 (3 months)

Built a metal defect classification pipeline using CNNs and improved model interpretability with Grad-CAM and LIME, reaching 99.67% accuracy. Enhanced regression performance to 97.36% by integrating PCA within the LMS framework and developed an LSTM Encoder–Decoder with Attention for seq2seq text summarization.

CNN Grad CAM LIME Model Interpretability Regression Modeling Attention Mechanisms Summarization