Keep in touch with meI'm using Intch to connect with new people. Use this link to open chat with me via Intch app
Work Background
Modern Streaming Solutions Private LimitedSoftware Developer
Dec. 2025Built a multimodal processing pipeline to analyze and index 500+ hours of unstructured telemetry data using YOLOv8, OpenCV, and LLM-based components, enabling searchable retrieval for auditing use cases.
Reduced data processing overhead by ~40% by implementing a multi-threaded extraction pipeline with frame-skipping logic, improving throughput for high-resolution video streams.
Developed validation workflows combining deep learning models with classical computer vision techniques (e.g., HSV thresholding) to support compliance-related checks.
Improved inference efficiency by applying model optimization techniques such as KV caching and quantization, reducing latency for multimodal query workloads.
ExternApplied Data Scientist
Oct. 2025 - Jan. 2026Tracy, California, United StatesBuilt modular, AI-powered pipelines to process 200+ page mortgage blob files—combining OCR (Tesseract, PaddleOCR), PDF parsing (PyMuPDF), and RAG techniques for intelligent data extraction, classification, and search.
Developed a document retrieval system using LlamaIndex and Retrieval-Augmented Generation (RAG), optimized for multi-document mortgage blobs. Enhanced precision through chunk tuning, metadata filtering, and evaluation of open-source LLMs like Mistral and Phi-2.
Conducted end-to-end evaluation of the document intelligence system on 200+ page mortgage blobs—benchmarking OCR accuracy, RAG retrieval quality, and routing performance. Delivered a technical report outlining model trade-offs, optimization strategies, final deployment recommendations, and built a UI for demo purposes.
ForageGlobal Markets Sales and Trading Analyst
Oct. 2025 - Oct. 2025Tracy, California, United StatesCompleted a job simulation focused on analyzing market trends and delivering client-centric solutions within the sales and trading division.
Conducted in-depth data analysis using tools like Excel and Bloomberg to identify key financial trends, assess market dynamics, and align insights with client objectives.
Researched and proposed strategic recommendations for optimizing trade execution processes and enhancing workflow efficiency using automation and process analysis.
Developed a client proposal outlining tailored investment strategies leveraging data-driven insights to address client goals such as portfolio diversification, sustainability, and moderate growth.
Triveni ITSoftware Developer
Aug. 2025 - Apr. 2026New Jersey, United StatesArchitected driver-based analytical pipelines integrating Oracle Fusion and Microsoft Dataverse to analyze financial market trends and optimize enterprise asset lifecycle management. Developed executive-facing Power BI and Tableau dashboards delivering real-time visibility into procurement KPIs, stock turnover, and profitability forecasting. Designed and deployed large-scale Robotic Process Automation (RPA) workflows to standardize global financial data ingestion, reducing manual processing risks and improving regulatory compliance.
Smart Rewards Inc.AI & ML Engineer
Aug. 2025New York, United StatesAutomating HR & Marketing Workflows: Build end-to-end pipelines on N8N to post job openings on CJNNow, YouTube, and LinkedIn automatically, reducing manual effort by ~60%.
Email Automation: Design and manage automated HR email campaigns, including candidate notifications and interview reminders, saving ~10–15 hours/week.
Social Media Content Management: Create, schedule, and publish posts on LinkedIn using ML-driven content optimization for higher engagement and relevance.
Workflow Reliability & Monitoring: Implement logging, error handling, and performance checks within N8N to ensure stable and repeatable operations.
Collaboration & Iteration: Partner with HR, marketing, and IT stakeholders to define requirements, validate workflow outputs, and continuously improve automation efficiency.
Tech & Tools: N8N, Python, REST APIs, YouTube API, LinkedIn API, Automation Workflows, Scheduling, Basic ML/NLP, Git, Documentation
Smart Rewards Inc.AI Intern
Jun. 2025 - Aug. 2025New York, United StatesDelivered a production RAG assistant that materially improved support KPIs: ~25% ticket deflection, ~70% reduction in manual search, and hours saved per agent per week in resolution time.
Drove adoption from a 20-person pilot to 4 departments (120+ DAUs; ~500 MAUs), handling 10,000+ queries/month with p95 <500 ms and 150+ concurrent users.
Full-stack ownership: corpus curation of ~5M internal docs, 768-d embeddings, ~50 GB Pinecone index; LangChain-based orchestration and evaluation harness for ongoing quality.
Productionized on GCP (Cloud Run + autoscaling), Dockerized services, and GitHub Actions CI/CD; implemented OpenTelemetry tracing and Grafana observability.
Reduced per-query costs by ~35% and nearly halved latency (p95 ~900 ms → ~500 ms) via caching, batching, and system-level performance tuning.
Increased trust and usage with citations, confidence scores, and feedback loops; led training and authored runbooks, driving 60% adoption lift; showcased by CTO and featured in customer-facing materials; live with roadmap-backed expansion.
Smart Rewards Inc.Data Science Intern
Jan. 2025 - Jun. 2025New York, United StatesScaled clinical analytics across ~6 Phase II/III trials (~2,500–3,000 patients, ~40 sites) by integrating ~20M+ records from EDC, CTMS, LIMS, and ePRO; daily/weekly data refreshes for operational currency.
Designed the analytics “product”: reusable Power BI templates and standardized ETL in Python/SQL Server that harmonized KPIs and accelerated new-study rollout.
Sole owner of enrollment and deviation monitoring dashboards with RLS for PHI segmentation; automated lab/ePRO pipelines with Git versioning and Airflow scheduling.
Embedded robust data governance: HIPAA and 21 CFR Part 11 compliance, de-identification, automated QC/anomaly detection, and cross-source reconciliation.
Delivered measurable operating leverage: ~60% reduction in manual work (~100 hours/month), ~3-day faster query resolution, and ~2-week improvement to database lock on two key trials.
Influenced study strategy via early SAE trend surfacing and variance analysis, reducing audit findings and regulatory exposure.
Drove adoption through bi‑weekly readouts to Clinical Ops/Biostats/Regulatory and periodic steerco updates; work highlighted at the R&D townhall to senior leadership.
ForageData Analyst
Nov. 2024 - Nov. 2024California, United StatesAnalyzed 100K–150K records from 3–5 structured and semi-structured sources to surface operational bottlenecks; recommendations enabled 10% cost reduction ($20K–$25K) via resource reallocation and process simplification.
Developed 3–4 Tableau/Power BI dashboards for operations leaders (≈20–25 active users) with week-1 MVP and full delivery in 3 weeks, improving time-to-insight and self-serve decision-making.
Built Python-based ETL pipelines (pandas, NumPy) with SQL integration and validation checks to consolidate ERP data, CSV exports, and operational logs; standardized pipelines improved data reliability and reduced manual effort by 50%.
Established and tuned KPI framework (turnaround time, completion rates, utilization) and performed anomaly detection and segmentation to target high-variance processes.
Partnered with operations and finance to translate findings into prioritized actions; presented readouts to manager-level stakeholders and ensured compliance with internal data governance (no PII/PHI exposure).
Drexel University College of Computing & InformaticsResearch Assistant
Oct. 2024 - Aug. 2025Philadelphia, Pennsylvania, United StatesDirected dataset curation and model development for a Drexel–Thomas Jefferson Hospital collaboration, training CLIP- and ViT-family models on ~10k+ mammograms from CBIS-DDSM and VinDr-Mammo under HIPAA-compliant governance (de-identification, DUAs, audit logs).
Engineered a dual-view fusion architecture tailored to mammography (CC/MLO) and tuned ViT/SWIN-T/CLIP backbones; achieved >90% internal accuracy with a 97% best-run on patient-level test splits; evaluated cross-site robustness.
Established rigorous ML practices (patient-level train/val/test, augmentation, hyperparameter sweeps) and MLOps (GitHub versioning, CI/CD, reproducibility); mentored a junior researcher on preprocessing and training standards.
Positioned findings for clinical relevance (early detection and triage efficiency) and academic visibility, with submissions in preparation for IEEE EMBC’25 and MICCAI’25.
Drexel University College of Computing & InformaticsResearch Intern
Sep. 2023 - Dec. 2023Philadelphia, Pennsylvania, United StatesOwned end-to-end modeling for CHD risk stratification on longitudinal EHR/claims data (~250K patients, 8 years, 120 features). Advanced AUC from ~0.74 (logistic) to ~0.89 with MAF/RealNVP; improved sensitivity by ~15% at >85% specificity, surfacing ~120 additional high-risk patients/month for earlier intervention.
Designed reproducible data/feature pipelines in Python/SQL on a secure Linux research cluster; harmonized Epic extracts; enforced IRB/HIPAA controls in a de-identified enclave; instituted run logging and validation checks for internal reproducibility.
Engineered domain signals (rolling labs, comorbidity index, refill gaps) and benchmarked classical ML (LogReg, RF, XGBoost) vs deep generative models; handled class imbalance with weighting + Platt scaling; used stratified/time-based validation, ablations, and SHAP review with clinicians.
Chick-fil-A RestaurantsTeam Member
Oct. 2022 - Jun. 2024Philadelphia, Pennsylvania, United StatesCoordinated peak-hour operations for a 10–12 person, campus-adjacent diner, handling 70–80 orders during lunch/dinner rushes and 30–50 orders per shift overall; regularly scheduled for the busiest blocks based on reliability and throughput.
Elevated guest satisfaction by 18% in 3–4 months by standardizing greetings, order read-backs, and handoff quality checks; served as point for peak-hour workflow coordination.
Cut order errors by 15% through a repeatable cashier→prep→bagging process, including read-backs, QC at handoff, and prep checklists—improving first-time-right accuracy.
Improved speed of service by ~10–15 seconds per order and increased throughput during peak windows by proactively reallocating staff and prioritizing high-velocity items.
Reduced waste by ~5% and minimized out-of-stocks via streamlined prep/replenishment routines and lightweight inventory tracking spreadsheets.
Trained and mentored 2–3 new hires over several months on POS operations, guest service standards, food safety basics, and quality control; occasionally entrusted with opening/closing, cash handling, and deployment.
Covered cashier/order taking, front counter, prep, bagging/expediting, inventory/weekly deliveries; maintained temp logs, hygiene compliance, and allergen-safe handling.
Applied graduate analytics/operations concepts to optimize workflows, prep prioritization, and error tracking during high-volume periods.
CognizantJunior Software Engineer
Jul. 2020 - Jul. 2022Bengaluru, Karnataka, IndiaFull-stack .NET IC owning feature delivery for three regulated pharma systems (SFC, JPUBS, ReCAP3), integrating .NET Framework, C#, SQL Server, JavaScript, and Azure DevOps to meet uptime and compliance expectations.
Scaled field-rep workflows in SFC to support ~200–300 active reps and ~1,500–2,000 daily call logs; upgraded validation and exception handling to reduce support tickets ~20% and elevate user satisfaction by 96% (internal survey).
Modernized JPUBS approval flows and resolved 9 high-severity issues to accelerate publication submissions for 50–100 users; improved reliability across 500–700 publications/quarter.
Delivered reusable components for ReCAP3 to automate routing and checks for ~50–80 proposals/week; upheld ~99% uptime during business hours and strengthened data quality.
Reliability and quality: Closed ~28 production defects; implemented repeatable testing and debugging practices in partnership with QA, improving deployment confidence and reducing rollbacks.
CI/CD and collaboration: Automated build/deploy pipelines in Azure DevOps; coordinated with client PMs and architects to validate requirements and secure approvals.
Intch is a Professional Networking App for the Future of WorkSee more people like Varun on Intch
300k+ people
130+ countries
AI matching
Get Intch if you'd like to
1107009 people
16
Owner/Operator @ Artistry & Hair Design by Jamie Lynn
16
Senior Strategist @ Freeman Company
15
CEO & Founder @ The Water Traveler
239205 people
16
Owner/Operator @ Artistry & Hair Design by Jamie Lynn
15
Project Manager @ Richland County School District One
17
Elementary Educator