● PORTFOLIO / V1.0
AI Engineer & Builder
DarshChandura
I'm an AI Engineer in Boston building production voice agents, RAG systems, and machine learning automation in Python. I've shipped systems handling $1M+ ad portfolios, 200,000+ products, and 45% efficiency gains across clean energy and HVAC markets.
What I Bring
Four areas where I ship production-grade AI work. Measured, deployed, and running in real markets.
Voice & Conversational AI
Production voice agents built with Synthflow, JotForm, and custom LLM integrations. I've shipped 5,000+ automated interactions that cut manual outreach time by 45% across clean-energy markets.
RAG & LLM Systems
Retrieval-augmented generation pipelines handling 12,000+ live queries. I've built RAG agents for student support and mental-health triage with A/B baselines and explicit kill criteria.
Data Pipelines & Web Scraping
Python scraping systems pulling structured data from 50+ vendor websites. The pipeline populated a marketplace with 200,000+ products and drove 38% transaction growth.
Marketing AI & Automation
Weather-based ad optimization across $1M+ portfolios. 22% cost-per-conversion drop, 12% ROI lift, all running through Zapier and custom-LLM workflows that generate SEO content.
About Me
Hi, I'm Darsh. I'm an AI Engineer in Boston, currently building production voice agents and LLM-powered automation at Modern Energy.
My work connects LLMs to real business outcomes. I've shipped RAG systems handling 12,000+ student queries at WPI, voice-based survey platforms used by 300+ people at Solomon Group, and Python scraping pipelines that put 200,000+ HVAC products into a marketplace. I care about measurement before conclusion. Every release I ship has a documented hypothesis, an A/B baseline, and a kill criterion.
I lead WPI's Data Science Council, a 30-person team running 15+ international technical events for 3,000+ students. WPI named me their GOAT of the Week in 2024 for AI innovation. Before WPI, I finished my Bachelor's in Information Technology at Mumbai University.
Featured Projects
Production AI systems I've shipped at Modern Energy, WPI, and Solomon Group. These are live or recently retired internal systems. Case study writeups are available on request.
VOICE AI / HVAC
HVAC Quote & Mapper
Instant-quote and contractor-mapping system that delivers quotes to homeowners in under 10 seconds. Built with Python, OpenAI, and Mapbox. The map visualizes 500+ contractor service areas across the Modern Energy network.
Python · OpenAI · Mapbox · Node.js · Vercel
DATA PIPELINES / SCRAPING
AI Web Scraping System
Python pipeline that uses supervised ML to extract and structure product data from 50+ HVAC vendor websites. Built with BeautifulSoup, Selenium, and Scrapy. Handles anti-bot detection, schema normalization, and automatic categorization. The output feeds a unified consumer marketplace.
Python · BeautifulSoup · Selenium · Scrapy · PostgreSQL · AWS Lambda
MARKETING AI / OPTIMIZATION
Weather-Based Ad Optimization
ML model that adjusts Google Ads and Meta budgets dynamically based on local weather forecasts. Built in Python with scikit-learn and the Google Ads API. Deployed across 5+ MassCEC clean-energy clients managing $1M+ in ad spend.
Python · scikit-learn · Google Ads API · Meta Marketing API · BigQuery
RAG / VOICE / EDUCATION
RAG Chatbot + Alexa Integration
Custom LLM with retrieval-augmented generation, plugged into Alexa AI Voice Assistant and a campus-wide chatbot. Built with Python, LangChain, LlamaIndex, and OpenAI. Handles student queries and makes campus resources easier to find at WPI.
Python · LangChain · LlamaIndex · OpenAI · Alexa Skills Kit
RAG / WELLBEING
Mental Health Triage Agent
AI agent for WPI's Center for Well-Being. It detects emotional cues, provides guided self-help responses, and connects students to on-duty counselors. Built in Python with LangChain and sentiment models. Ships with structured response-quality measurement and explicit kill criteria.
Python · LangChain · Custom LLM · Sentiment Models · Firebase
VOICE AI / CLEAN ENERGY
Mass Save Voice-Based Energy Survey
Converted Mass Save's energy assessment survey into a voice-based experience using Domain-Adaptive Fine-Tuning (DAFT) on Alexa devices. Built with Python, AWS Lambda, and DynamoDB. Homeowners complete the survey hands-free and get personalized solar recommendations.
Python · DAFT · Alexa AI · AWS Lambda · DynamoDB
Skills & Tooling
What I work with day-to-day.
PROGRAMMING & SCRIPTING
Programming & Scripting
AI & MACHINE LEARNING
AI & Machine Learning
DATA, CLOUD & INFRASTRUCTURE
Data, Cloud & Infrastructure
MARKETING, CRM & AUTOMATION
Marketing, CRM & Automation
Work Experience
Four years building production AI systems across clean energy, education, and analytics.
AI Engineer
CURRENTModern Energy · USA · Remote
- ·Deployed AI agents (JotForm, Synthflow) for voice-based cold calling and customer service. 5,000+ automated interactions, 45% reduction in manual outreach time.
- ·Built marketing automation linking QuickBase, Google Ads, and DUDA via Notion AI + Zapier. 20% conversion lift, 37% engagement growth.
- ·Shipped HVAC Quote & Mapper systems delivering quotes in <10s to 1,200+ homeowners and visualizing 500+ contractor locations. 28% conversion-to-booking lift, sales cycle 14→6 days.
- ·Built AI web-scraping system extracting structured product data from 50+ HVAC vendor websites. Populated marketplace with 200,000+ products driving 38% transaction growth.
- ·Developed weather-based ad-optimization model adjusting Google/Meta budgets dynamically. 22% lower CPC, 12% ROI lift across $1M+ portfolio for 5+ MassCEC clean-energy clients.
Data Analyst
Worcester State University · USA
- ·Built automated AI-driven analytics system in Python + Salesforce CRM tracking enrollment, retention, and academic performance. 23% lift in qualified lead generation.
- ·Shipped “EduIntel360,” a student-facing platform on QuickBase enabling real-time tracking of academic progress, attendance, and insights. Adopted by 5,000+ students.
- ·Platform contributed to a 12% enrollment lift the following term.
Data Analyst
Worcester Polytechnic Institute · USA
- ·Developed custom LLM with RAG, integrated into Alexa AI and a campus-wide chatbot for student queries. 19% reduction in help-desk requests.
- ·Built mental-health support chatbot for WPI's Center for Well-Being detecting emotional cues, providing guided responses, and routing students to counselors. 3,000+ interactions; 27% drop in reported mental health incidents through early intervention.
- ·Created AI-powered content marketing workflow using Zapier + custom LLM. 39% organic traffic lift, 28% impression growth across digital platforms.
Data Scientist
Solomon Group · USA
- ·Converted Mass Save Energy Assessment Survey into a voice-based Alexa experience using DAFT and voice-recognition algorithms. Homeowners complete energy assessments hands-free and receive personalized solar recommendations.
- ·Adopted by 300+ users in first month. 34% lift in survey completion, 18% lift in qualified lead generation, 25% improvement in energy optimization accuracy.
Educational Background
Graduate work in Data Science with a concentration in AI, building on undergraduate IT foundations.
Master's Degree in Data Science
Concentration: Artificial Intelligence
Worcester Polytechnic Institute · Worcester, MA
Graduate work focused on applied AI: RAG systems, LLM fine-tuning, voice agents, and production ML. My capstone work spanned conversational AI and educational analytics.
Bachelor of Engineering in Information Technology
Concentration: Data Science
Universal College of Engineering, Mumbai University · Mumbai, IN
Undergraduate foundation in IT and data science with a strong emphasis on algorithms, databases, and applied programming.
Awards & Recognition
Honors received for AI innovation, hackathons, competitive programming, and public speaking.
● FEATURED RECOGNITION
GOAT Award 2024
Worcester Polytechnic University
September 2024
WPI named me Greatest Of All Time of the Week for AI innovation. The recognition cited the RAG-powered Alexa chatbot that handles 12,000+ student queries and the mental health triage agent at the Center for Well-Being that contributed to a 27% drop in reported incidents.
CodeShastra 6.0 Winner
National Hackathon, India
First place in CodeShastra 6.0, a national-level hackathon, for an AI solution shipped within the competition timeframe.
AlgoBulls Winner
National Algorithmic Trading Competition
First place for a quantitative trading strategy with the highest risk-adjusted returns in the national-level AlgoBulls competition.
National Public Speaking Award
National Level Public Speaking Competition
Recognized for technical communication and presentation skills at a national-level public speaking competition.
Speaking & Workshops
Teaching what I build. Invited talks and workshops on AI, measurement, and production LLM systems.
● INVITED WORKSHOP
Data Detectives
AI Workshop at the STEM Expo
Worcester State University
October 2024
Worcester State invited me to lead a hands-on AI workshop at their STEM Expo. I taught faculty, students, and the local community how to build measurement frameworks around AI experiments. We covered hypothesis design, A/B baselines, kill criteria, and the difference between a demo and a deployed system.
- Designing AI experiments with documented hypotheses
- A/B testing frameworks for LLM-powered workflows
- Kill criteria: knowing when to roll back a model
- Measurement before conclusion: case studies from WPI and Modern Energy
Brands I've Worked With
A selection of brands and organizations I've worked with across roles at Modern Energy, Worcester Polytechnic Institute, FCB Kinnect, Buildingease, HP Marketing Pros, and others.
Let's Work Together
Have a project in mind? Drop a line. I'd love to hear about it.