AI Implementation Engineer
As an AI Implementation Engineer, you take a customer from a business problem to an AI model running on their production line. You start with the customer's problem, not the data: what is hurting them, what result would prove that AI solves it, and what data is needed to demonstrate that. You then build the model using our software and present the results to the customer.
You do not write the model architecture. The quality of the model depends on your judgment: which images you select, how you define each defect, and how you interpret the model's mistakes.
The role is half technical and half communication-focused. You will own customer projects end to end, running proof-of-value projects for new customers and keeping live models healthy for existing ones. It suits someone who thinks clearly, speaks and writes well, is patient with data, and enjoys seeing their work used in a real factory.
Key Responsibilities
Discover the Customer's Problem
Understand Pain Points: Talk to customer engineers and managers to understand their challenges, such as defects escaping to their customers, good dies being rejected due to overly stringent inspection, or the need for a single model that works across many devices.
Define Success Criteria: Turn each pain point into a clear, measurable target agreed upon with the customer.
Plan Data Requirements: Identify and request the data needed to demonstrate success.
Define Defect Classes: Agree on what each defect looks like, why it matters, and which defects must never be missed. Create clear labelling guidelines.
Build and Test the Model
Select Training Samples: Understand how AI learns and choose representative examples. A few hundred well-chosen images are better than thousands of random ones.
Improve Data Quality: Correct inaccurate labels and ensure there are sufficient examples for every defect type.
Train AI Models: Use the SixSense AI platform to train models and run structured experiments.
Analyse Model Errors: Review incorrect predictions image by image to identify incorrect labels, missing examples, unclear definitions, or model limitations.
Validate Model Performance: Test on unseen data against agreed success criteria, rather than relying solely on overall accuracy.
Measure Key Metrics: Analyse confusion matrices, accuracy, over-rejection, under-rejection, and automation rates.
Deploy and Monitor: Launch models on customer servers or in the cloud in collaboration with data science and software teams, and monitor their live performance.
Automate with Code: Use SQL and Python to extract data, validate results, and automate repetitive tasks.
Report and Present Results
Report Customer Impact: Build reports around the customer's pain points and agreed success criteria.
Present Model Results: Explain results clearly to customer engineers and managers, building trust and driving action.
Address Tough Questions: Handle challenging discussions calmly and communicate model limitations honestly.
Own Customer Communication: Serve as the primary technical contact and keep internal teams updated.
Drive Product Improvements: Bring customer issues and feature requests to the product and engineering teams.
Qualifications
Must Have
Excellent Communication: Strong written and spoken communication skills. You can explain technical findings without jargon and present confidently to both engineers and managers.
Customer Handling: You ask the right questions to get to the root of a problem and are comfortable leading technical discussions with customers.
Strong Logical Thinking: You can break a complex problem into steps, form a hypothesis, and test it.
Education: A bachelor's degree in engineering (any branch).
Experience: 3–5 years of work experience.
Familiarity with Programming: Basic knowledge of Python or another programming language. You can write and read simple scripts.
Familiarity with SQL: Basic knowledge of SQL. You can write simple queries to retrieve and validate data.
Patience and Attention to Detail: You are comfortable reviewing thousands of images and identifying subtle differences.
Ownership: You see a project through from the first conversation to production, and customers can rely on you.
Good to Have
Basic knowledge of machine learning or deep learning, including training, validation, overfitting, and class imbalance.
Exposure to computer vision or image data.
Experience in semiconductors, electronics, manufacturing, or quality inspection.
Familiarity with quality and yield metrics.
Experience preparing reports and presenting them to customers.
Other Details
You should be open to travelling to customer sites as required by projects.
You may need to work with customers across different geographies.
Full training on our product and the semiconductor domain will be provided.
Get to know us better through our culture, tech updates, and life at SixSense on our LinkedIn page.
Join us to engineer the intelligence behind every chip the world will ever make. Be part of our innovative and passionate team.
Skills Required
- Fast learner and problem solver who can apply new concepts independently
- Detail-oriented and patient when reviewing thousands of images and identifying subtle issues
- Clear communicator able to explain findings to customers and teammates
- Committed to owning projects from start to finish
- Available to start immediately or with minimal notice
- Willingness to travel occasionally to customer sites
- Hands-on experience or basic understanding of computer vision or AI models
- No specific degree or prior AI experience required; full training provided
What We Do
Founded by alumni from IIT and NIT, and funded by top investors in Asia, SixSense is a Singapore and India based startup in the AI space. We have built a no-code platform that allows users to build, monitor and maintain deep learning models for computer vision applications. Currently our platform is primarily used by manufacturing engineers, to build AI models that can detect and classify defect in images. Using SixSense AI platform, they are transitioning from their decade old process of inspection using manual labour or rule-based systems, to a much efficient and scalable AI based technology.
