Company Description

Moleculyst Ventures Private Limited is an AI startup building software for detecting and analyzing hallucinations in large language models (LLMs) .

Our work focuses on understanding when and why language models generate unreliable outputs and developing systems that can identify these failures during model inference. We work with open-weight language models and analyze model outputs, internal computations, and inference-time signals to build reliable hallucination detection and monitoring systems.

The broader goal is to build infrastructure that makes AI systems more reliable, observable, and trustworthy for real-world deployment.

Role Description

We are looking for a Python Developer to help build the software infrastructure behind our hallucination detection platform.

You will work on systems for running language models, capturing and processing inference-time data, building data pipelines, implementing backend services, and storing and analyzing large volumes of model-generated information.

You will collaborate closely with AI/ML researchers to turn experimental methods into reliable and scalable software.

Responsibilities

- Design, build, and maintain clean, modular, and testable Python systems.
- Build infrastructure for running and evaluating open-weight large language models.
- Develop pipelines for capturing, processing, filtering, and storing inference-time data.
- Build APIs and backend services for model evaluation and hallucination detection.
- Work with large numerical datasets, model activations, tensors, metadata, prompts, and generated outputs.
- Integrate databases and storage systems for efficient retrieval and analysis.
- Write unit and integration tests to ensure correctness and reproducibility.
- Profile and optimize performance-critical Python code.




- Collaborate with AI/ML researchers to convert experimental research code into robust software.
- Document system architecture, APIs, data formats, and engineering decisions.

Qualifications

- Strong Python programming skills.
- Strong understanding of object-oriented programming, data structures, and software design.
- Ability to write clean, modular, maintainable, and well-tested code.
- Experience with NumPy and PyTorch .
- Familiarity with backend development and building APIs.
- Experience with SQL and/or NoSQL databases.
- Understanding of Git, testing, debugging, and standard software-development practices.
- Experience processing large or complex datasets.
- Comfortable working with Linux and command-line development environments.
- Ability to independently investigate technical problems and implement solutions.
- Solid communication skills and ability to work effectively in a remote team.

Good to Have

- Experience with PyTorch internals or transformer-based language models .
- Familiarity with Hugging Face Transformers or other open-weight LLM tooling.
- Experience running or serving LLMs.
- Understanding of GPU computing, CUDA, or performance optimization.
- Experience with Docker and cloud infrastructure.
- Familiarity with model inference, forward hooks, activations, attention, KV caches, or other transformer internals.
- Experience building ML experimentation or evaluation infrastructure.

What You Will Work On A typical system you may help build looks like: Prompt → LLM Inference → Internal/Output Data Capture → Signal Processing → Hallucination Detection → Storage/API The role sits at the intersection of Python engineering, ML infrastructure, and LLM interpretability/reliability .

- This is a full-time remote role requiring ownership, independent problem-solving, and close collaboration with the research team.

📌 Python Developer (India)
🏢 Moleculyst
📍 India