Careers
We're a small, focused team building private AI memory and contextual intelligence for small and medium businesses. If you are passionate about maths, memory, and machines — and curious about what AI hasn't done yet — we'd like to meet you.
Upcoming Positions
Research Scientist — Deep Memory & Cognitive AI
Full-time · Remote-friendly · PhD required
Achiral AI is building shared memory infrastructure for teams — persistent, structured, semantically searchable memory that compounds over time across tools and people. We believe the next meaningful leap in that system comes from a deeper understanding of how human memory actually works: retrieval, forgetting, salience, habit, goal-directed attention, and learned procedural patterns.
We're looking for a researcher with a PhD in AI/ML, Cognitive Neuroscience, Computational Neuroscience, or a closely related discipline who wants to translate insights from memory science into applied AI architecture. This is an exploratory, high-autonomy role — you'll shape what we build, not just execute on a spec.
What you'll work on
- Research and prototype memory models inspired by cognitive architectures, automaticity, conscious access, and memory retrieval — and evaluate their viability in production AI systems
- Design retrieval and encoding strategies that go beyond surface-level semantic search — associative retrieval, temporal decay, source weighting, and task-context signals
- Collaborate with the engineering team to integrate research findings into Chiro, our AI assistant, and the underlying vector memory layer (Weaviate + vLLM)
- Publish and contribute to the public discourse on memory-augmented AI — this is work we're proud of and want known
- Define and run evaluations that measure memory quality, not just retrieval accuracy
What we're looking for
- PhD in AI/ML, Cognitive Neuroscience, Computational Neuroscience, or equivalent research experience
- Genuine curiosity about memory, automaticity, retrieval, and conscious access — whether approached from a neuroscience, cognitive science, or machine learning angle
- Comfort moving between theory and implementation — you can write a research proposal and a working prototype in the same week
- Familiarity with mathematical models of human memory — ACT-R activation and decay equations, Bayesian forgetting curves, or connectionist approaches — and an instinct for how those translate into retrieval system design
- Familiarity with transformer architectures, RAG systems, or vector databases is a strong plus — but we care more about your mental model than your stack
- A publication record or demonstrable research output in a relevant domain
AI Research Engineer — Language Models & Memory Systems
Full-time · Remote-friendly · PhD preferred
Achiral AI runs a self-hosted, multi-tenant inference fleet on NVIDIA DGX Spark hardware — fine-tuned models, LoRA adapters, a two-tier vLLM serving setup, and a Weaviate-backed memory layer that compounds across thousands of organisations. The system works. We want someone to make it significantly better.
This is a research engineering role for someone who moves fluidly between reading papers and shipping production code. You'll own the ML layer end-to-end — from training runs and adapter design to serving optimisations and retrieval quality improvements. A PhD in ML, Computer Science, or a related quantitative field is strongly preferred; an exceptional research track record without one is equally welcome.
What you'll work on
- Design and train LoRA adapters for domain-specific fine-tuning across organisation tenants — improving response quality without compromising the shared inference fleet
- Improve retrieval quality in the RAG pipeline — hybrid BM25/vector search, re-ranking, context compression, and evaluation harnesses that go beyond BLEU and recall@k
- Optimise the vLLM serving stack — quantisation, batching strategies, KV-cache tuning, and latency/throughput tradeoffs at the per-tenant quota level
- Collaborate with the memory research team to translate cognitive memory models into concrete retrieval and encoding mechanisms inside the vector layer
- Define the evaluation framework for the AI system as a whole — not just whether it retrieves the right chunk, but whether it makes the team measurably smarter
What we're looking for
- PhD in Machine Learning, Computer Science, Statistics, or a related quantitative discipline — or equivalent depth demonstrated through research output
- Hands-on experience with transformer architectures, parameter-efficient fine-tuning (LoRA, QLoRA, adapters), and the tradeoffs involved in deploying them at scale
- Solid Python and familiarity with at least one of: vLLM, Hugging Face Transformers, LlamaIndex, LangChain, or comparable inference/orchestration tooling
- Experience designing and running ML evaluations — you have strong opinions about what good benchmarks look like and why most of them are wrong
- Comfort working close to hardware — GPU memory constraints, precision formats, and throughput arithmetic are not abstract concepts to you
- Bonus: experience with vector databases, multi-tenant ML systems, or knowledge graph construction
Whisperer
Full-time · Remote-friendly
A Whisperer at Achiral is a developer who builds by directing AI — precisely, intentionally, and with craft. Not a user of AI tools. A practitioner who knows how to decompose a problem, frame it for a model, and get production-quality output on the first or second attempt. You write less code than you used to. You ship more than you ever have.
This role exists because we believe the gap between an average developer and a great one is closing — but the gap between someone who can wield AI fluently and someone who can't is opening fast. We want the former.
What you'll work on
- Build features across the Achiral stack (Next.js + Express + MongoDB + Weaviate) using AI-assisted development as your primary working mode
- Translate product requirements directly into working, tested code — with AI doing the heavy lifting where appropriate and you steering the outcome
- Debug, review, and refine AI-generated code with the same rigour you'd apply to code you wrote yourself — you own the output
- Help establish and document prompting patterns that work for our codebase, so the whole team benefits from what you learn
What we're looking for
- 1–3 years of software development experience, with a genuine fluency in AI-assisted coding (Cursor, Copilot, Warp, Claude, or equivalent — pick your weapon)
- Strong enough fundamentals to catch what the model gets wrong — you understand the code you ship, even when you didn't write every line
- Familiarity with JavaScript/TypeScript, React, and REST APIs
- A portfolio, repo, or project you can walk us through that was built significantly with AI assistance — we want to see your prompting instincts in action
Sr. Whisperer
Full-time · Remote-friendly
The Sr. Whisperer has been here before — the late nights reading model docs, the prompts that almost worked, the moment something clicked and velocity tripled. You've moved past treating AI as an autocomplete and into treating it as a collaborator with a strong opinion and specific weaknesses you know how to work around.
At Achiral you'll own entire product surfaces, mentor Whisperers, and help us define what engineering looks like when AI does most of the implementation work and humans do most of the thinking.
What you'll work on
- Lead end-to-end delivery of complex features — architecture, implementation, and review — using AI-assisted development as the default mode of work
- Define the prompting patterns, context-loading strategies, and agentic workflows the whole team adopts — you set the standard
- Mentor Whisperers: code review with an eye on how AI was used, not just what it produced
- Push back on product scope when the constraint is genuinely AI capability, not engineering effort — you know the difference
- Contribute to Chiro — you'll use the thing you're building, and that makes your feedback unusually valuable
What we're looking for
- 5+ years of software development experience, with demonstrable mastery of AI-assisted workflows at scale — not just productivity hacks, but a repeatable method
- Deep JavaScript/TypeScript, Node.js, and React experience; comfortable with distributed systems and API design
- Experience mentoring junior developers and raising the quality bar on a team
- A strong point of view on where AI-assisted development is going and the intellectual honesty to update it when you're wrong
- Bonus: experience with LLM integrations, RAG pipelines, or vector search
Shepherd
Contractor · Remote · Independent
When a new organisation signs up to Achiral, a Shepherd is assigned to them from day one. You guide them through provisioning, help them get real value from Chiro quickly, and stay close as their AI memory grows. You're the human layer that makes the platform work for people who aren't engineers.
This is a contractor role designed to be built into a practice. The more organisations you shepherd well, the more your reputation compounds — much like the memory system you'll be helping them build. Shepherds who do this well can grow a meaningful independent book of business around it.
No two clients are the same. You'll need enough technical fluency to understand what Chiro is doing and enough people instinct to translate it into something a 10-person accounting firm or a busy legal team can actually act on.
Don't see a role that fits? We read speculative applications. Email [email protected] with a note about what you'd want to build.