Senior Machine Learning Engineer

thalolabs · New York, NY

seniorfull timeonsite
Apply on thalolabs’s site →

Role Snapshot

A senior individual-contributor role at an early-stage physical-AI startup in Midtown Manhattan. You will ship ML, statistical, and LLM-based fault-detection models on proprietary HVAC sensor data, build the AI eval harness from the ground up, and translate model outputs into insights field teams can act on. Salary: $150K–$180K, heavily equity-weighted. Fully onsite. Visa sponsorship not confirmed in the posting.

About This Role

Thalo Labs is hiring a Senior Machine Learning Engineer to lead the intelligence layer of its HVAC monitoring platform. The company captures proprietary streaming sensor data from heat pumps and commercial HVAC equipment — a dataset no outside organization has access to — and this role is responsible for making sense of it. Working end-to-end, from raw time-series ingestion through to customer-facing features, you will define how the platform detects equipment faults, surfaces anomalies, and generates trustworthy recommendations that field technicians and building operators can act on without second-guessing.

About Thalo Labs

Thalo Labs is building a physical AI layer for the HVAC industry. Its premise: the global push toward heat pumps and electrification will put over 100 million new units into service this decade, but the service workforce and maintenance infrastructure are nowhere near ready to support them. Thalo's answer is sensors plus physics-grounded intelligence that convert static equipment into self-diagnosing systems, shifting service from reactive guesswork to data-driven precision. The founding team brings experience from Waymo, Google, NASA, John Deere, and Boom Supersonic — and is applying that same engineering discipline to one of the most consequential infrastructure buildouts of the coming decade.

What You'll Own

  • Lead the development and continuous improvement of the fault-detection engine covering electrical, refrigerant, and equipment-performance diagnostics
  • Research, prototype, and deploy ML, statistical, and LLM-driven models against proprietary streaming HVAC sensor time-series data
  • Build and operate the AI evaluation infrastructure: curate labeled fault datasets from real service outcomes and field cross-checks, define precision and recall metrics, and maintain a regression harness that gates every model change or prompt update before it reaches production
  • Convert complex model outputs into concise, actionable reports that field technicians, customer success, and business development teams can present confidently to customers
  • Continuously strengthen the data pipeline — ingestion, storage, transformation, and analysis — so detection scales reliably and cost-effectively as sensor coverage grows
  • Collaborate across hardware, software, and business functions to close the loop between field insights and the product roadmap, and document your work so the entire team can build on it

Must-Have Requirements

  • 5+ years of experience building and deploying ML or statistical models on production data, ideally within an early-stage startup environment
  • Graduate degree (M.S. or Ph.D.) in mathematics, physics, statistics, data science, or a comparable quantitative field — or demonstrably equivalent applied experience
  • Hands-on production track record with time-series or streaming sensor data across areas such as anomaly detection, forecasting, or signal processing
  • Proven experience shipping features on frontier LLMs in production — including prompt engineering, structured outputs, tool use, and RAG — with clear judgment on when an LLM is the right choice versus a deterministic rule or statistical model
  • Experience designing and running AI evaluations: building labeled eval sets, measuring precision and recall, using LLM-as-judge, and detecting regressions as prompts and underlying models change over time
  • Strong Python fluency and software engineering practices sufficient to deliver production-grade code, not just exploratory notebooks
  • Ability to translate model outputs into plain-language insights that a field technician or building operator will trust and act on without hesitation
  • Genuine curiosity about how physical systems behave and a drive to understand the reasoning behind product decisions, not just the implementation
  • Self-directed ownership mindset with a consistent habit of documenting decisions and sharing context so teammates can build on your work

Bonus Points

  • Genuine enthusiasm for sustainability and the global energy transition
  • Domain knowledge in HVAC, refrigeration, combustion, building systems, or energy — Thalo is willing to help the right candidate develop this expertise on the job
  • Experience with agentic or tool-using LLM systems, RAG over technical documentation, or LLM vision models
  • Familiarity with LLM efficiency techniques such as prompt caching, batch inference, output drift monitoring, and model governance practices including A/B-testing context changes
  • Production deployment of frontier or open-source LLMs, or hands-on use of AWS Bedrock
  • Full-stack capability to carry a feature through to a React/TypeScript UI; experience with time-series databases (InfluxDB, TimescaleDB) and dashboarding tools such as Grafana

Skills at a Glance: Required vs. Nice to Have

The non-negotiables are Python, production ML and statistics, and direct experience with time-series or streaming sensor data. LLM skills — prompt engineering, RAG, tool use — and AI evaluation design are equally required, not optional. The nice-to-haves expand the role's footprint: React/TypeScript for UI work, InfluxDB and TimescaleDB for time-series storage, Grafana for observability, and AWS (including Bedrock) for infrastructure. HVAC domain knowledge is a genuine advantage but is explicitly listed as teachable — engineering depth and product judgment carry more weight than industry pedigree.

Salary & Equity

The employer-stated base salary range is $150,000–$180,000 per year. Thalo Labs explicitly frames its compensation structure as equity-weighted, meaning a meaningful share of total compensation is expected to come from equity at this early-stage company. The posting does not disclose the equity percentage, vesting schedule, or cliff — candidates should request those specifics during the interview process. No salary figure outside the stated $150K–$180K range has been published; JobForMe does not estimate or adjust compensation beyond what the employer has disclosed.

Benefits & Perks

  • Subsidized national healthcare coverage for medical, dental, and vision
  • 401(k) retirement savings plan
  • Twelve weeks of paid parental leave
  • Paid time off
  • Complimentary mental health support and professional coaching sessions through Lyra
  • Weekly team social hours and quarterly company off-sites
  • Stocked pantry at the Midtown Manhattan office

Location & Work Arrangement

This position is fully onsite at Thalo Labs' office in Midtown Manhattan, New York, NY. The posting does not reference remote or hybrid flexibility. Candidates should be based in New York City or prepared to relocate. Visa sponsorship and relocation assistance are not addressed in the posting — applicants who need either should raise the question directly with the company before applying.

Frequently asked questions

Is this a remote or hybrid role?

No. The posting specifies a fully onsite position at Thalo Labs' Midtown Manhattan office. Remote or hybrid arrangements are not mentioned in the job description.

Does Thalo Labs offer visa sponsorship?

The posting does not mention visa sponsorship in either direction. Candidates who require work authorization assistance should clarify directly with the employer before applying.

Is HVAC domain knowledge required to apply?

No. Thalo lists HVAC, refrigeration, and building-systems experience as a bonus and explicitly states it is willing to help the right candidate learn this domain. Core ML engineering, LLM, and time-series skills are the genuine requirements.

What does the equity component look like?

The posting states that Thalo places significant emphasis on equity at this early stage but does not disclose the specific percentage, cliff, or vesting schedule. Candidates should request those details during the interview process.

What kind of data will I be working with?

The role works with proprietary streaming HVAC sensor time-series data that Thalo describes as unique — not academic benchmarks or publicly available datasets. Part of the role also involves building labeled fault datasets from real service outcomes, physics-versus-LLM disagreements, and field cross-checks.

Are you a fit for this role?

See how your background matches Senior Machine Learning Engineer — and the 20 live jobs that fit you best. No signup, no email, nothing stored.