Raj Ranpariya

HVAC comfort advisor and project manager who builds AI systems.

Built to learn, verified against the running system

Projects I built to learn and to demonstrate what I can do, mostly in the problem space I know best: field-service operations.

Every one was built through directed AI-assisted implementation: I defined the problem, the workflow, the gates and the acceptance criteria, directed the implementation, and verified the result against the running system. Descriptions say what runs today and what is only designed.

Most cards link to GitHub: design notes, diagrams and the real verification harnesses, with product code kept private.

Phone screen: adding a contractor's quote by photo, paste or voice

Phone screen: the deposit limit in Massachusetts

Quote check and contractor finder for homeowners

Problem: A homeowner holding a contractor’s quote cannot tell what is missing or whether the terms are fair.
What it does: Reads a quote from a photo, a PDF or typed text and flags what is missing or risky: a deposit above the state’s limit, no cancellation notice, no license number, no finish date. It also finds licensed contractors from public license records.

The rule: It suggests and never decides, and every message to a contractor is sent by the homeowner. The state checks come from a written rule file covering all 50 states and DC, not from a model’s memory.
Evidence: Tested on 20 sample quotes with 60 planted problems; a live check flagged a 50% deposit against Massachusetts’ one-third limit.

An assistant for a homeowner trying to judge a contractor’s quote. It reads the quote from a photo, a PDF or typed text and flags what is missing or risky, such as a deposit above the state’s limit, no cancellation notice, no license number or no finish date. It finds licensed contractors from public license records, and its state checks come from a written rule file for all 50 states and DC, with deposit limits where a state sets one. It suggests and never decides, and any message to a contractor is sent by the homeowner.

Field app with a 3D crew packet

Problem: The sales visit, the proposal and the install handoff spread across paper forms and separate tools.
What it does: One offline file for the iPad: the sales comparison, the proposal, rebates and the install handoff, with the original survey, 3D model and crew packet still inside. The handoff fills equipment sizes from manufacturer spec tables, dates existing units from their serial numbers and measures from a LiDAR room scan opened on the iPad.

The rule: Every value it fills shows the document and page it came from, and it says plainly what still blocks “ready for install” instead of guessing.
Evidence: The page’s security policy blocks every network request, it uses no AI service, and customer files never leave the device.

A single-file iPad app I first built for my own day job, grown into one offline tool for the whole job: the sales comparison, the proposal, rebates and the install handoff behind one header. The handoff fills equipment sizes from manufacturer spec tables, each value with its document and page, dates existing units from their serial numbers, measures from a LiDAR room scan opened on the iPad itself, and says plainly what still blocks “ready for install” before it prints the installer package. The original survey flow, with its live 3D model and crew packet, is still inside. It makes no network request, uses no AI service, and customer files never leave the device.

On GitHubDesign notes + diagram

Field-service agent system

119 tests14/14 daily smoke

Problem: A live caller will not wait, an agent that books real teams can overwrite people, and permits are unforgiving of a missed step.
What it does: A seven-assistant voice squad triages calls and hands off to six specialists; other agents book and reschedule, turn a won job into a work order a manager reviews, track permits and inspections as an explicit state machine, and handle invoices, reviews and referrals.

The rule: A booking exists only when a row is written; installation, inspection, callback and emergency bookings are held for human approval; nothing is ever auto-submitted; two hardcoded invariants no configuration can weaken sit beneath every outbound message.
Evidence: A daily automated check exercises 14 agents’ write paths and the last seven runs passed 14 of 14; the safety layer has 119 tests, and two independent adversarial review rounds found 25 issues before deployment.

One system built around a field-service job, from the first call to the follow-up. The voice platform allows 30 seconds per tool call, so conversational work runs in-process and slower work goes to a workflow engine behind a 25-second abort; caller identity comes from telephony metadata, never from the model. A weekly text-simulation harness runs three scripted scenarios across 18 trades; the latest run passed 18 of 18. Done means an external effect a separate checker verified: deleting each gate in turn found four sub-checks removable with zero failures, which now assert behaviour. What is not live yet is said plainly: dispatch recommendations remain stubs, the headless-browser sidecar is built and health-checked but no workflow calls it yet, and the 48-hour and seven-day follow-up sequence is designed and has not yet run.

On GitHubDesign notes + diagram Test excerpt (29 of 119)Fleet smoke harness

Comparison and presentation tool for the sales conversation

Problem: A customer feels comfort, efficiency and sound, not specifications.
What it does: One self-contained HTML file for a tablet: photos, drawings, calibrated sound demos, a comfort-and-performance ladder, plain-language explainers and a one-tap PDF summary.

The rule: Built by a salesperson for the sales conversation; it works offline and travels by AirDrop.
Evidence: The sound demos are calibrated so the differences you hear are proportionally real.

A presentation tool for comparing heating and cooling systems on the three things a customer actually feels: comfort, efficiency and sound. One self-contained HTML file for a tablet, with embedded photos, drawings and unit sound demos calibrated so the differences you hear are proportionally real, a comfort-and-performance ladder that builds as you add systems, plain-language explainers on right-sizing, humidity and warranties, and a one-tap PDF summary the customer can take away. Built by a salesperson for the sales conversation; it works offline and travels by AirDrop.

On GitHubDesign notes + diagram

Backorder Buster

Problem: A unit that is sold, scheduled and then turns out to be backordered.
What it does: Checks the availability of every model on the upcoming installs and classifies each as in stock, low stock, backordered or restricted.

The rule: A short emailed report lands before the install date is at risk, while a swap or a warning is still possible.
Evidence: Built because I was the person who had to make that phone call.

Backorder Buster, a small automation for my day job that attacks the most expensive surprise in a sales pipeline: a unit that is sold, scheduled and then turns out to be backordered. It signs in to the distributor portal with a captured session, checks the availability of every model on the upcoming installs, classifies each one as in stock, low stock, backordered or restricted, and emails a short report before the install date is at risk, so a model can be swapped or a customer warned while there is still time. Built because I was the person who had to make that phone call.

On GitHubDesign notes + diagram

Years inside the workflow before automating it

I know this work from the inside.

About six years in customer-facing technical roles in heating and cooling, as a comfort advisor and project manager: discovery with homeowners and business owners, heating-load and system design, financing conversations, permits, installation coordination, technician ride-alongs and inspections.

Since June 2024 I have sold $7.79M of heating and cooling across 1,046 sales opportunities, $3.34M of it in 2025, and my close rate has climbed from 37% in the second half of 2024 to 48% in 2025 and 53% in 2026 through October. Every one of those jobs then had to be designed, permitted, installed and inspected under my name. That is where the judgment in my projects comes from: I know which step gets skipped, which call gets missed, and what a customer needs to hear before they say yes.

Close rate

the second half of 202437%
202548%
2026 through October53%

Deterministic where the input is structured. A model only where judgment is needed. Humans at the one-way doors.

I define the problem, the workflow, the acceptance criteria and the human gates; I direct AI to build it, then verify the result against the running system rather than trusting a report that says it works.

agent
human gate / external check
operator

Where the humans are is the design. The model handles judgment; every one-way door has a gate.

Reducing the friction between uncertainty and action

The same method outside work.

I use my own judgment plus AI in every area where the friction between uncertainty and action is high, and the mechanism is always the same: the real documents and the real question go in, a structured checklist or comparison comes out, and every line is read by me before anything is filed, chosen or acted on. The model does not decide and does not file.

Market monitor and scheduled accumulation

A personal market-monitoring and scheduled-accumulation agent: eleven scheduled jobs on a small Linux server, running since April 2026.

Deterministic scripts score a five-component market regime daily from free public data, apply a long-term trend filter to a watchlist, snapshot a paper portfolio, and write a graded weekly review to my phone. Anything touching real money executes only after an approval reply from me; an unanswered request expires after two hours and places nothing. A file-based kill switch is checked at every point on the order path, and the self-heal loop may restart monitoring jobs but never the money path.

Life admin: taxes, benefits, first-time parent

Personal administration, the same way: a tax-season paperwork pass, a benefits-enrollment comparison, and the research that comes with being a first-time parent.

Source documents and the actual question in; a checklist, a comparison table, or a list of questions to ask a professional out; every line reviewed by me before anything is filed or chosen. The value is not the generated text. It is shortening the distance between an unfamiliar problem and a reviewed, concrete next step.

Health routine and vegetarian meal planning

Health and diet as a running project: a standing assistant that holds my routine and plans a vegetarian week around it, adjusting when travel or work breaks the pattern.

Nothing clinical, no numbers to chase; just a consistent plan that survives contact with a real schedule.

A private second brain, an MCP layer, spoken briefings

A private second brain I built for my own work.

Postgres with pgvector holds verbatim evidence chunks and extracted memories, every row carrying provenance and a sensitivity level. Local open-weight models on one small server embed and extract nightly under a CPU cap, and a Markdown wiki is regenerated from the database each night. Eleven MCP tools expose search and recall to my coding assistant behind a per-client sensitivity ceiling. A publication firewall, a claims ledger, an approval queue and a deterministic leak gate wired into the site build, is the only path from private notes to this site. A spoken-briefing path to my phone is prototyped.

On GitHubSource: site, firewall, assistant, Jarvis

I sell heating and cooling systems and I build AI systems, and each one makes me better at the other.

In sales I earn a customer's yes and then own the project until the inspection passes. At the desk I build agents and check them against running systems, not against reports that say they work. The sales work teaches me what a customer actually needs to hear; the engineering teaches me what a promise really costs. This site is a worked example of the same habit in every part of life: my own judgment plus artificial intelligence, applied to sales, projects, engineering, money, health and the paperwork of being a person.

About

Customer-facing technical sales for homes and commercial sites, project ownership from the signed proposal to the final inspection, and AI systems built by directing AI rather than hand-writing code: that is the combination I bring. I use AI to get more out of every part of life, professional and personal, and I verify what I build against running systems rather than trusting a report that says it works.

I sell and manage heating and cooling projects for a living, for homes and commercial sites, and in 2026 I started building software seriously: agents, orchestration, and the verification around them, first as personal tools and then as a full multi-agent system built as an independent project.

I build with AI as an engineering partner. I define the problem, the workflow, the acceptance criteria and the human gates; I direct the implementation, then verify it against the running system rather than trusting a report that says it worked.

What I can actually do

Sales and customer-facing

Discovery with homeowners and business owners, consultative selling, financing conversations, objection handling, closing, post-sale ownership until the inspection passes.

Project management

Heating and cooling load analysis, system design and right-sizing, permits and inspections, installation coordination, technician ride-alongs, crews, inspectors and municipalities.

HVAC and building science

Test-and-balance, airflow, combustion safety, home-performance diagnostics, retrofit and new construction, light-commercial systems.

AI systems, built by direction

I do not hand-write code. I define the problem, the workflow, the acceptance criteria and the human gates; I direct AI to build it, then verify the result against the running system. Agent orchestration, voice-to-action, human-in-the-loop gates, evaluation harnesses, retrieval with provenance.

Hosting and infrastructure

Linux servers and Docker, databases and vector stores, local open-weight models, DNS, domains and edge hosting, firewalls and backups, webhooks and API integrations: chosen, set up and kept running.

Email is the door

Email is the door: [email protected]. If you are building or deploying AI where real customers and real operations are involved, I would like to hear about it.

Email me Résumé (PDF)GitHub

Talk to my second brain

Prefer to ask instead of read? The assistant below answers questions about my work using only the text on this page: what a project does, why it matters, what I bring to a role. It will not discuss my personal life or how anything was built; for that, write to me.

Answers come only from the text on this page. Personal questions and build details get a polite no and my email: [email protected].

Raj Ranpariya

Quote check and contractor finder for homeowners

Field app with a 3D crew packet

Field-service agent system

Comparison and presentation tool for the sales conversation

Backorder Buster

Years inside the workflow before automating it

Deterministic where the input is structured. A model only where judgment is needed. Humans at the one-way doors.

Market monitor and scheduled accumulation

Life admin: taxes, benefits, first-time parent

Health routine and vegetarian meal planning

A private second brain, an MCP layer, spoken briefings

Customer understanding, technical sales, domain workflow, AI systems, orchestration, verification

About

What I can actually do

Sales and customer-facing

Project management

HVAC and building science

AI systems, built by direction

Hosting and infrastructure

Email is the door

Talk to my second brain