Improbable
At Kallikor, we're building the future of supply chain intelligence through AI-powered simulation digital twins. We create living digital representations of real-world operations (warehouses, distribution networks, global logistics) that help organisations make better decisions faster. We're at an inflection point: moving from AI-assisted tools to domain-specific AI that understands supply chains as deeply as our best engineers do. You'll be instrumental in building our first domain-specific language model (DSML) and the foundation for Project Genome, an ambitious initiative to capture and synthesize the world's supply chain knowledge into actionable intelligence. This is a production engineering role first. You'll build robust Python systems that happen to train and serve LLMs, not the other way around. We need someone who writes production-quality code, debugs complex distributed systems, and thinks about reliability, who has learned ML/LLMs as powerful tools in their engineering arsenal. You'll work across our entire AI stack: building FastAPI services that serve models, creating training pipelines that process production data, deploying inference endpoints with proper monitoring, and integrating all of this into our existing Python backend. The ML is important, but the engineering discipline is what makes it production-ready. Learn more at kallikor.ai. Your Opportunity Build production AI systems: Design and implement the full stack, from FastAPI endpoints that handle requests, to training pipelines that process data, to inference services that serve predictions. You'll own the architecture, not just the model weights. Train and deploy our DSLM: Fine-tune models using Unsloth/Axolotl, but more importantly, build the robust infrastructure around it - data pipelines that feed training, evaluation frameworks that catch regressions, deployment systems that handle failover. Make it production-grade. Integrate ML into our backend: We use FastAPI, PydanticAI, FastMCP, Memgraph. You'll extend these systems with ML capabilities, not as a separate "ML service" but as a natural part of our backend architecture. Clean abstractions, proper error handling, observability. Own inference performance: Get models running fast, whether that's vLLM deployment, quantization strategies, batching optimizations, or caching. Hit our Shape Project Genome's foundation: Work with our Principal Engineer to architect how we ingest, process, and learn from global supply chain data. This is systems design as much as ML with data pipelines, graph databases, incremental learning strategies being just as important. Mentor through code review and pairing: Raise the bar on code quality, testing, and production practices across the team. Teach mid and junior engineers how to build ML systems that don't fall over. Why you're made for this You're a strong production Python engineer: You write clean, maintainable, tested code. You understand async/await, know when to use generators vs lists, can profile performance bottlenecks. You've built FastAPI services (or similar) that handle production traffic. Your code passes review without drama. You've built with LLMs in production: You've integrated GPT-4/Claude into real applications, handled streaming responses, dealt with rate limits and retries, cached intelligently. You know the practical challenges: prompt engineering, context management, error handling, cost control. You've trained or fine-tuned models: Whether it's fine-tuning LLMs, training classifiers, or running experiments, you understand the workflow. You've dealt with training data quality, evaluation metrics, and overfitting. You can debug why a model isn't learning what you expected. You think like a systems engineer: You design for failure, add instrumentation, consider edge cases. You know that "the model works on my laptop" isn't shipping. You care about monitoring, logging, alerting, and graceful degradation. You can navigate the ML landscape pragmatically: You know enough about transformers, attention mechanisms, and training dynamics to make informed decisions. But you're not precious about it. If a simple heuristic beats a complex model, you ship the heuristic. You balance velocity with quality: You ship incrementally and iterate based on production data. But you don't accumulate tech debt, you refactor proactively, write tests that matter, and leave the codebase better than you found it. You communicate trade-offs clearly: You can explain to the team why we're choosing LoRA over full fine-tuning, why we're deploying on Fireworks instead of self-hosting, or why a 7B model might beat a 70B model. You help everyone make informed decisions. What we're looking for specifically Must have: 5+ years building production Python systems (backend services, APIs, data processing) Strong software engineering fundamentals: design patterns, testing, debugging, profiling Experience integrating LLMs into applications (OpenAI/Anthropic APIs, prompt engineering, streaming, PydanticAI) Understanding of ML training workflows (even if you're not an expert. You need to know enough to build the infrastructure) Docker, CI/CD, production deployment experience Can read and understand PyTorch code (you don't need to write novel architectures) Nice to have: Fine-tuning experience (LoRA, full fine-tuning, QLoRA) Distributed training basics (DeepSpeed, FSDP) Graph databases (Memgraph, Neo4j) Supply chain or logistics domain knowledge Experience with agent frameworks (LangChain, PydanticAI, etc.) What you'll work with Backend Stack: Python, FastAPI, PydanticAI, FastMCP, Memgraph, Postgres ML Stack: PyTorch, Unsloth/Axolotl for training, vLLM for inference, Weights & Biases Models: Qwen 2.5, Llama 3.1, GPT-4, Claude (for now) Infrastructure: AWS (flexible), Docker, Kubernetes, GPUs when needed Team: Principal Engineer (your partner on architecture), Mid Data/ML Engineer (your data pipeline partner), Junior AI Engineer (your mentee) Example projects you'll own Build a FastAPI service that handles streaming LLM responses with correct error handling and retry logic Create a training pipeline that processes production logs, validates data quality, and triggers fine-tuning runs Deploy a fine-tuned 7B model with vLLM that beats GPT-4 latency while maintaining quality on our domain Design the data ingestion architecture for Project Genome, how we process papers, documentation, and operational data at scale Implement evaluation frameworks that catch model regressions before they reach production About Us Kallikor is determined to foster an environment where people can do their best work and feel like they belong. We believe a healthy culture, strong values and contribution from a diverse range of individuals will help us to achieve success. We do not discriminate based on race, ethnicity, gender, ancestry, national origin, religion, sex, sexual orientation, gender identity, age, disability, veteran status, genetic information, marital status or any other legally protected status. Improbable Candidate Privacy Policy
At Kallikor, we're building the future of supply chain intelligence through AI-powered simulation digital twins. We create living digital representations of real-world operations (warehouses, distribution networks, global logistics) that help organisations make better decisions faster. We're at an inflection point: moving from AI-assisted tools to domain-specific AI that understands supply chains as deeply as our best engineers do. You'll be instrumental in building our first domain-specific language model (DSML) and the foundation for Project Genome, an ambitious initiative to capture and synthesize the world's supply chain knowledge into actionable intelligence. This is a production engineering role first. You'll build robust Python systems that happen to train and serve LLMs, not the other way around. We need someone who writes production-quality code, debugs complex distributed systems, and thinks about reliability, who has learned ML/LLMs as powerful tools in their engineering arsenal. You'll work across our entire AI stack: building FastAPI services that serve models, creating training pipelines that process production data, deploying inference endpoints with proper monitoring, and integrating all of this into our existing Python backend. The ML is important, but the engineering discipline is what makes it production-ready. Learn more at kallikor.ai. Your Opportunity Build production AI systems: Design and implement the full stack, from FastAPI endpoints that handle requests, to training pipelines that process data, to inference services that serve predictions. You'll own the architecture, not just the model weights. Train and deploy our DSLM: Fine-tune models using Unsloth/Axolotl, but more importantly, build the robust infrastructure around it - data pipelines that feed training, evaluation frameworks that catch regressions, deployment systems that handle failover. Make it production-grade. Integrate ML into our backend: We use FastAPI, PydanticAI, FastMCP, Memgraph. You'll extend these systems with ML capabilities, not as a separate "ML service" but as a natural part of our backend architecture. Clean abstractions, proper error handling, observability. Own inference performance: Get models running fast, whether that's vLLM deployment, quantization strategies, batching optimizations, or caching. Hit our Shape Project Genome's foundation: Work with our Principal Engineer to architect how we ingest, process, and learn from global supply chain data. This is systems design as much as ML with data pipelines, graph databases, incremental learning strategies being just as important. Mentor through code review and pairing: Raise the bar on code quality, testing, and production practices across the team. Teach mid and junior engineers how to build ML systems that don't fall over. Why you're made for this You're a strong production Python engineer: You write clean, maintainable, tested code. You understand async/await, know when to use generators vs lists, can profile performance bottlenecks. You've built FastAPI services (or similar) that handle production traffic. Your code passes review without drama. You've built with LLMs in production: You've integrated GPT-4/Claude into real applications, handled streaming responses, dealt with rate limits and retries, cached intelligently. You know the practical challenges: prompt engineering, context management, error handling, cost control. You've trained or fine-tuned models: Whether it's fine-tuning LLMs, training classifiers, or running experiments, you understand the workflow. You've dealt with training data quality, evaluation metrics, and overfitting. You can debug why a model isn't learning what you expected. You think like a systems engineer: You design for failure, add instrumentation, consider edge cases. You know that "the model works on my laptop" isn't shipping. You care about monitoring, logging, alerting, and graceful degradation. You can navigate the ML landscape pragmatically: You know enough about transformers, attention mechanisms, and training dynamics to make informed decisions. But you're not precious about it. If a simple heuristic beats a complex model, you ship the heuristic. You balance velocity with quality: You ship incrementally and iterate based on production data. But you don't accumulate tech debt, you refactor proactively, write tests that matter, and leave the codebase better than you found it. You communicate trade-offs clearly: You can explain to the team why we're choosing LoRA over full fine-tuning, why we're deploying on Fireworks instead of self-hosting, or why a 7B model might beat a 70B model. You help everyone make informed decisions. What we're looking for specifically Must have: 5+ years building production Python systems (backend services, APIs, data processing) Strong software engineering fundamentals: design patterns, testing, debugging, profiling Experience integrating LLMs into applications (OpenAI/Anthropic APIs, prompt engineering, streaming, PydanticAI) Understanding of ML training workflows (even if you're not an expert. You need to know enough to build the infrastructure) Docker, CI/CD, production deployment experience Can read and understand PyTorch code (you don't need to write novel architectures) Nice to have: Fine-tuning experience (LoRA, full fine-tuning, QLoRA) Distributed training basics (DeepSpeed, FSDP) Graph databases (Memgraph, Neo4j) Supply chain or logistics domain knowledge Experience with agent frameworks (LangChain, PydanticAI, etc.) What you'll work with Backend Stack: Python, FastAPI, PydanticAI, FastMCP, Memgraph, Postgres ML Stack: PyTorch, Unsloth/Axolotl for training, vLLM for inference, Weights & Biases Models: Qwen 2.5, Llama 3.1, GPT-4, Claude (for now) Infrastructure: AWS (flexible), Docker, Kubernetes, GPUs when needed Team: Principal Engineer (your partner on architecture), Mid Data/ML Engineer (your data pipeline partner), Junior AI Engineer (your mentee) Example projects you'll own Build a FastAPI service that handles streaming LLM responses with correct error handling and retry logic Create a training pipeline that processes production logs, validates data quality, and triggers fine-tuning runs Deploy a fine-tuned 7B model with vLLM that beats GPT-4 latency while maintaining quality on our domain Design the data ingestion architecture for Project Genome, how we process papers, documentation, and operational data at scale Implement evaluation frameworks that catch model regressions before they reach production About Us Kallikor is determined to foster an environment where people can do their best work and feel like they belong. We believe a healthy culture, strong values and contribution from a diverse range of individuals will help us to achieve success. We do not discriminate based on race, ethnicity, gender, ancestry, national origin, religion, sex, sexual orientation, gender identity, age, disability, veteran status, genetic information, marital status or any other legally protected status. Improbable Candidate Privacy Policy
Improbable
At Improbable, we build and scale ambitious AI and Web3 ventures shaping the next generation of the internet. We're hiring for a venture building a next-generation blockchain for real-time, large-scale applications, designed to overcome the speed and cost limitations of traditional blockchain infrastructure, the technology is fast, cost-efficient and fully EVM-compatible, enabling developers, creators and communities to build immersive products with greater ownership and interoperability. About the Role We're hiring an Engineer to turn bold ideas into working demos that show developers what's possible. This is a hands-on role for someone who loves building and shipping fast. You'll take concepts across DeFi, trading systems and agentic workflows, and turn them into live demos, prototypes and technical content that bring the technology to life and inspire developers to start building. You'll think independently, move quickly and create experiences that make complex technology feel clear, useful and exciting. If you use AI tools to accelerate how you think, build and deliver, you'll thrive in this role. What you'll be doing You'll create the demos, prototypes and technical content that help developers and partners see what potentially can be built. Build demo dApps that bring the technology to life, from real-time trading interfaces and onchain games to agent-powered applications Create trading bot demos that highlight execution quality, onchain order books and new interaction models Prototype agentic applications, including autonomous agents coordinating onchain, prediction market resolvers and agent-to-agent workflows Produce video walkthroughs, live demos and coding content that make complex ideas easy to understand and exciting to explore Develop reusable starter kits, templates and reference projects that help developers start building faster Spot ecosystem trends early and turn them into timely, compelling demo ideas Partner with DevRel, Marketing and BD to support launches, partnerships and broader go-to-market activity What we're looking for Strong Solidity and EVM development experience, including writing, deploying and interacting with smart contracts Ability to build polished frontend interfaces for demos and prototypes A track record of turning ideas into working products, prototypes or experiments Strong instincts for what makes a demo memorable Comfortable working at speed and shipping rough but effective prototypes in days, not weeks High conviction, independent thinking and a bias toward action AI-native working habits, using tools to accelerate research, ideation, building and delivery This role is ideal for someone who loves building in public, experimenting quickly and creating the proof points that get developers, partners and the wider ecosystem excited. Apply if you want to build the demos people talk about.
At Improbable, we build and scale ambitious AI and Web3 ventures shaping the next generation of the internet. We're hiring for a venture building a next-generation blockchain for real-time, large-scale applications, designed to overcome the speed and cost limitations of traditional blockchain infrastructure, the technology is fast, cost-efficient and fully EVM-compatible, enabling developers, creators and communities to build immersive products with greater ownership and interoperability. About the Role We're hiring an Engineer to turn bold ideas into working demos that show developers what's possible. This is a hands-on role for someone who loves building and shipping fast. You'll take concepts across DeFi, trading systems and agentic workflows, and turn them into live demos, prototypes and technical content that bring the technology to life and inspire developers to start building. You'll think independently, move quickly and create experiences that make complex technology feel clear, useful and exciting. If you use AI tools to accelerate how you think, build and deliver, you'll thrive in this role. What you'll be doing You'll create the demos, prototypes and technical content that help developers and partners see what potentially can be built. Build demo dApps that bring the technology to life, from real-time trading interfaces and onchain games to agent-powered applications Create trading bot demos that highlight execution quality, onchain order books and new interaction models Prototype agentic applications, including autonomous agents coordinating onchain, prediction market resolvers and agent-to-agent workflows Produce video walkthroughs, live demos and coding content that make complex ideas easy to understand and exciting to explore Develop reusable starter kits, templates and reference projects that help developers start building faster Spot ecosystem trends early and turn them into timely, compelling demo ideas Partner with DevRel, Marketing and BD to support launches, partnerships and broader go-to-market activity What we're looking for Strong Solidity and EVM development experience, including writing, deploying and interacting with smart contracts Ability to build polished frontend interfaces for demos and prototypes A track record of turning ideas into working products, prototypes or experiments Strong instincts for what makes a demo memorable Comfortable working at speed and shipping rough but effective prototypes in days, not weeks High conviction, independent thinking and a bias toward action AI-native working habits, using tools to accelerate research, ideation, building and delivery This role is ideal for someone who loves building in public, experimenting quickly and creating the proof points that get developers, partners and the wider ecosystem excited. Apply if you want to build the demos people talk about.
