AI/ML Engineer Intern

What matters most

• Build and deploy conversational AI

  The posting explicitly calls for building and deploying AI chatbots and virtual assistants. Prepare to explain how a Python-based prototype becomes a usable assistant, including how you would check its responses.

• Optimize prompts across LLMs

  Optimizing prompt engineering for various LLM models is a stated responsibility, and LLM familiarity is required. Prepare a concrete example of comparing prompts and judging whether a change improved the output.

• Develop document-processing pipelines

  The role includes document extraction and processing. Prepare to describe the stages of a pipeline and how you would verify that extracted information is accurate enough for its intended use.

• Apply ML foundations to new technology

  The posting asks for Python and ML knowledge, experience with TensorFlow, PyTorch or a similar framework, and research and integration of emerging AI/ML technologies. Prepare to explain how you would assess a new tool before incorporating it into a solution.

Likely challenges to think through

• Balancing assistant usefulness and reliability

  Inferred preparation hypothesis, not a stated employer problem: a chatbot may give fluent answers that are wrong or unhelpful. Prepare to discuss the trade-off between broad responses and controlled behavior, using examples of checks you would run before deployment.

• Handling variation across models

  Inferred preparation hypothesis: a prompt that works well with one LLM may perform differently with another. Prepare a comparison method that keeps the task and evaluation criteria consistent while testing model-specific prompt changes.

• Processing inconsistent documents

  Inferred preparation hypothesis: differing document layouts or missing fields could make extraction brittle. Prepare to explain how you would detect failures and decide when to improve the pipeline rather than accept lower extraction accuracy.

Topics to prepare

• Machine learning fundamentals

  Review training versus inference, overfitting, and evaluation. Exercise: explain how you would split example data and choose a metric for an extraction task. Relevance: the posting explicitly requires machine learning concepts and analytical problem-solving.

• LLMs and prompt engineering

  Review how instructions, examples, and output constraints affect an LLM’s response. Exercise: compare two prompts for the same assistant request and explain which errors each might produce. Relevance: the role requires LLM familiarity and prompt optimization across models.

• Chatbots and autonomous agents

  Review the distinction between a conversational assistant and an agent that takes actions. Exercise: sketch a request-to-response flow and identify where errors should be checked. Relevance: the role describes autonomous agents and building virtual assistants; the flow is a preparation suggestion, not a specified architecture.

• Document extraction pipeline design

  Review how source documents become extracted fields and how uncertain or missing values can be handled. Exercise: diagram the stages for extracting fields from a sample document. Relevance: developing document extraction and processing pipelines is an explicit responsibility.

Skills to revise

• Python implementation

  Exercise: write a small Python function that reads sample text, extracts named fields, and handles missing values. Relevance: a strong Python background is explicitly required.

• Prompt testing

  Exercise: draft two prompts for one chatbot task, test them against the same sample requests, and record failure cases; compare models if available. Relevance: the role calls for optimizing prompts for various LLMs.

• AI framework practice

  Exercise: use TensorFlow, PyTorch, or a similar framework you know to run a small training or inference example and explain the output. Relevance: experience with an AI framework is explicitly required.

• Assistant prototyping

  Exercise: build a minimal Python assistant that accepts a user request, returns a response, and handles an unsuccessful response gracefully. Relevance: building and deploying chatbots and virtual assistants is a stated responsibility; this prototype is a revision exercise, not a specified deliverable.

• Document processing implementation

  Exercise: process a few sample documents into consistent field records and inspect incorrect or missing extractions. Relevance: the posting explicitly calls for document extraction and processing pipelines.

• Prepare a Python and ML example

  The posting asks for Python, machine-learning concepts, and experience with TensorFlow, PyTorch, or a similar framework. Prepare a short walkthrough of work you have done, including the problem, your implementation, and how you evaluated it.

• Sketch a chatbot deployment

  Building and deploying chatbots and virtual assistants is a stated responsibility. As a design exercise, diagram a user request flowing through a model to a response, and note how you would test behavior and handle failures.

• Show how you evaluate prompts

  The role includes optimizing prompts for different LLMs. Prepare a small comparison with sample prompts, test inputs, outputs, and a clear reason for any revision.

• Walk through a document pipeline

  Document extraction and processing are named responsibilities. Prepare a project walkthrough or design sketch covering input documents, extraction, validation, and handling of missing or incorrect fields.

• Explain a technology choice

  The role includes researching and integrating emerging AI/ML technologies. Choose one relevant tool or approach and prepare a brief assessment of its use case, limitations, and how you would test it before integration.

• Draft role-specific questions and an application statement

  Prepare questions about the chatbot, document-processing, and autonomous-agent work, including how an intern would contribute. The application form asks why Cynbit, so draft a concise statement grounded in those responsibilities rather than assumed projects.