AI Engineer
Build production retrieval, structured scoping, and evaluation pipelines that transform human briefs into scoped projects in milliseconds.
Takes about 3 minutes · Direct human review · No cover letter required
Job Snapshot
- Department
- Engineering
- Location
- Remote (India)
- Employment Type
- Full-time
- Experience Level
- 2–5 yrs
- Work Model
- Remote-first (India)
- Practice
- Core Product Delivery
The Opportunity
Overview & Strategic Impact
As an AI Engineer at Pixirain, you will tackle the foundational challenge of our business: translating an unstructured, conversational paragraph from a non-technical founder into a structured, executable project scope with guaranteed pricing and clear deliverables. You will build, benchmark, and deploy automated extraction and evaluation pipelines that operate with deterministic accuracy and sub-second latency.
About the Team & Practice
Pixirain AI focuses exclusively on high-utility, evaluated intelligence. We don't build gimmicky chatbots. Our systems parse unstructured human project briefs, perform semantic scoping, evaluate package suitability, and connect clients to the exact technical specialists capable of executing their vision with mathematical precision.
What You'll Do & Deliver
Concrete responsibilities you will own from your first month onward:
Design and maintain the brief-to-package classification and structured scoping pipeline using modern LLM APIs, embedding spaces, and retrieval systems.
Construct comprehensive, automated evaluation sets (Evals) to measure precision, recall, and hallucination rates across thousands of edge cases.
Implement robust schema validation, output repair, and guardrail layers ensuring downstream systems only receive strictly validated JSON schemas.
Benchmark models across speed, cost, and accuracy, knowing precisely when a simple deterministic rule or heuristic outperforms a model.
Monitor live production inference pipelines, track token economics, latency drifts, and model regression over time.
Work with full-stack engineers to embed intelligent suggestions, brief clarifiers, and automatic package matchers into the live user interface.
What We Expect & How Success is Measured
Transparent milestone goals for your first 30, 60, and 90 days:
- 30 Days:Analyze current matching error logs, build your first automated eval dataset, and ship an improvement to the brief scoping parser.
- 60 Days:Deploy an end-to-end pipeline upgrade that increases matching precision by 15% while reducing token inference cost.
- 90 Days:Own the roadmap for intelligent project breakdown, milestone prediction, and automated deliverable checklist generation.
- Ongoing:Maintain an empirical, benchmark-driven approach where no prompt or model change reaches production without passing test suites.
Skills & Technical Toolkit
Tools, languages, and technical competencies central to this position:
What We're Looking For
- 2+ years of production software engineering experience, including at least one shipped LLM/AI feature used by real paying customers.
- Demonstrated expertise in building rigorous evaluation harnesses rather than relying on qualitative spot checks.
- Deep familiarity with state-of-the-art model families (Claude, GPT, Gemini, open-source models via Ollama/vLLM).
- Strong engineering fundamentals in data structures, algorithms, and distributed asynchronous workflows.
Nice to have (bonus, not mandatory)
- Experience fine-tuning models (LoRA, QLoRA) on domain-specific corpora
- Published articles, benchmarks, or open-source repositories demonstrating novel AI agents or eval systems
- Experience in talent matching, search engines, or e-commerce recommendation algorithms
Perks & Comprehensive Benefits
We invest deeply in our people so they can do their life's best work:
High-tier competitive salary and substantial founding-team equity package.
Uncapped API budget for research, benchmarking, and exploratory model experimentation.
Remote work independence with flexible hours centered around deliverables, not seat time.
Top-spec workstation stipend (Apple Silicon MacBook Pro or high-end Linux machine).
Comprehensive family healthcare coverage and mental wellness stipends.
₹75,000 annual learning grant for attending technical conferences and AI research seminars.
Our Selection Process
We respect your time. Every stage is purposeful and structured:
Application Review
A person on the team reads your submission within 48 hours.
Introductory Conversation
30 min video chat covering what you've built, your goals, and team fit.
Craft Deep Dive
60 min deep dive into one architecture, codebase, or case study you owned.
Paid Working Session
A real, self-contained backlog task at your standard consulting rate.
Offer & Welcome
We decide within three working days with transparent compensation & equity.
Ready to build with us?
Four fields, zero red tape, and reviewed by the people who lead the work.