AI/ML Engineer

Location: IIT Kanpur |

Type: Full-time |

Level: Mid-level

About Invictron Invictron Systems develops autonomous defense technologies that operate reliably in GNSS-denied, GPS-contested, and electronically hostile environments. The company builds end-to-end autonomous warfare solutions, including intelligent ground control systems, UAV software, and real-time object detection, tracking, and aiming to enhance situational awareness and decision-making. Founded and led by IIT alumni and mentors, Invictron focuses on creating sovereign battlefield autonomy for India, reducing reliance on imported drone and UAV systems. While its primary focus is defense and aerospace, Invictron also aims to adapt these technologies for commercial applications, driving broader real-world impact. We're a small team. Everyone here works close to the hardware and the mission.

The Role We are not looking for someone who fine-tunes a pretrained model and considers the job finished. We are looking for someone who builds the model: makes the architecture decisions, owns the training pipeline, and identifies performance bottlenecks before anyone else notices them. You will take a mission requirement, translate it into a loss function, and proactively suggest improvements. That is the core of the role. Part of the job is to train, optimize, and deploy on real hardware. You will work closely with the robotics integration engineer, since the model's accuracy in a notebook is not the standard here. It has to perform inside the actual system, on the actual compute, at the actual frame rate. You will be responsible for reviewing simulated performance data and determining whether the model is ready for flight, and if it is not, identifying exactly what needs to change. We also expect someone who can explain the model's behavior in plain terms to a non-technical stakeholder, then debug the integration queries with a robotics engineer. Both kinds of conversation are part of this role.

Requirements

• Solid hands-on experience training and building AI/ML models from scratch, not just using off-the-shelf ones
• Strong grounding in computer vision
• Edge AI Deployment, understanding of computer architecture well enough to optimize models for real hardware constraints, not just accuracy on a benchmark
• Systems thinking: you can see how a model decision affects the pipeline it sits inside, not just the model itself
• Strong analytical skills and comfortable working with strategic or non-strategic requirements down to a concrete training objective
• Able to communicate technical work to both technical and non-technical audiences
• Open to relocation as the team scales -

How to Apply To qualify, fill out this Google Form: https://docs.google.com/forms/d/e/1FAIpQLSffucI5HgXhXlkxmQdODJS919tuP6MwEYLt63XADzHknxDyQg/viewform?usp\=publish\-editor #AIML #ComputerVision #ModelTraining #CUDA #DeepLearning #DroneEngineering #Hiring #MachineLearning #EdgeAI