Job: AI Engineer (LLMs + Applied ML) — 2+ years experience
Torinit| AI/ML-first work | Build real systems, not demos
India | Remote
Overview
Torinit Technologies Inc. is a Canadian-based digital consulting company. At Torinit, we don't just serve our clients; we work with them to create transformative digital journeys by leveraging the latest technologies and world-class best practices. Our goal is to create an environment of continuous learning and success for our team and our clients.
We are a fast-growing team of passionate problem-solvers and life-long learners who aren't afraid of tackling complex problems. We've got a talented team of skilled, friendly, and driven individuals spanning the globe (primarily based in Canada and India), now we're looking to add more world-class talent to our team.
We're looking for someone with at least 2 years of experience who's excited about LLMs + applied AI and wants to grow fast while building production-grade AI features with a strong engineering team around them.
This is not a you must know everything role. If you've done a few things well, and you're hungry to learn the rest, you'll fit in.
This role requires availability from 2:00 PM to 11:00 PM IST to support cross-team collaboration.
What you'll work on (high level)
- Build AI capabilities from initial experimentation through production deployment and monitoring.
- Develop multimodal document-intelligence workflows using OCR, vision models, LLMs and deterministic processing.
- Extract structured information from documents, tables, drawings and unstructured business records.
- Design retrieval pipelines using chunking, metadata filtering, hybrid search, reranking and grounded generation.
- Build structured-output workflows using schemas, validation, retries, fallbacks and confidence checks.
- Develop controlled agentic workflows using tool calling, function calling and multi-step orchestration.
- Create evaluation datasets, quality metrics and regression tests for probabilistic systems.
- Investigate hallucinations, extraction failures, retrieval misses and edge cases using structured failure analysis.
- Optimise model selection, prompts, context, latency, token usage and infrastructure cost.
- Build APIs, asynchronous workers, queues, event-driven workflows and streaming interfaces.
- Design human-review workflows for cases where AI confidence is insufficient.
- Implement tracing, monitoring, cost tracking and production observability.
- Work with product managers, engineers and domain experts to translate requirements into measurable AI outcomes.
- Contribute reusable components, architecture patterns and engineering standards for Torinit's AI practice
What we expect you to already know (must have)
- Software engineering
- At least two years of professional software-development experience.
- Strong programming skills in Python.
- Experience building production APIs and backend services.
- Experience with REST APIs, third-party integrations and asynchronous processing.
- Strong understanding of relational databases and data modelling.
- Familiarity with NoSQL databases where appropriate.
- Experience writing clean, maintainable and testable code.
- Understanding of error handling, logging and production debugging.
- Familiarity with Git and modern software-development workflows.
- Experience deploying or supporting applications in a cloud environment.
Experience with JavaScript or TypeScript is valuable, but Python will be the primary language for most AI workflows.
- Applied AI and LLM systems
- Integrating LLM APIs into applications
- Prompt and context design
- Structured generation using JSON schemas or typed outputs
- Embeddings and semantic search
- Retrieval-augmented generation
- Document processing and information extraction
- Tool and function calling
- Model-output validation
- Retry and fallback strategies
- AI evaluation
- Hallucination and grounding risks
- Model latency, token usage and cost management
You should be able to explain:
- What you built
- What you personally owned
- Where the system failed
- How you evaluated its quality
- Which trade-offs you considered
- How you improved the system
- What you would do differently now
What distinguishes a strong candidate
Strong candidates will demonstrate depth in at least one area, such as:
- Taking an AI capability from experimentation to production
- Building and improving a RAG, document-intelligence or extraction workflow
- Diagnosing AI failures and improving accuracy, latency, cost or reliability
- Creating evaluation frameworks, meaningful technical projects or production backend systems
We value engineering judgement, measurable outcomes and ownership more than familiarity with a long list of frameworks.
Preferred experience
Experience with some of the following is valuable:
- Cloud AI: Amazon Bedrock, Textract, Azure AI Document Intelligence, Google Document AI, OpenAI, Anthropic or Gemini
- AWS development: Hands-on experience using AWS SDKs such as Boto3 or the AWS SDK for JavaScript/TypeScript to integrate services including Bedrock, Textract, S3, Lambda, SQS and ECS
- RAG and orchestration: LangGraph, LangChain, LlamaIndex, vector databases, hybrid search and reranking
- Document AI and ML: OCR, layout analysis, multimodal models, table extraction, computer vision and information extraction
- Backend and infrastructure: FastAPI, Docker, serverless systems, queues, caching, streaming, observability and automated AI evaluation
Experience with Amazon Bedrock, Textract and AWS SDK-based integrations is preferred, but strong experience with another major cloud AI platform is also acceptable.
What success looks like:
During your first 30 days
You will:
- Understand our AI systems, engineering standards and active customer problems.
- Contribute to an existing production AI workflow.
- Learn the relevant datasets, evaluation methods and known failure modes.
- Ship a meaningful improvement to an existing service or pipeline.
During your first 90 days
You will:
- Own a clearly defined AI capability or workflow.
- Establish or improve an evaluation baseline.
- Diagnose and resolve important production failure cases.
- Improve at least one measurable dimension such as quality, latency, cost or reliability.
- Contribute to technical design and architectural decisions.
Within six months
You will:
- Own a clearly defined AI capability or workflow.
- Establish or improve an evaluation baseline.
- Diagnose and resolve important production failure cases.
- Improve at least one measurable dimension such as quality, latency, cost or reliability.
- Contribute to technical design and architectural decisions.
The kind of person who wins here
- High ownership: You do not stop after implementing the happy path. You investigate failures, close gaps and remain accountable for production outcomes
- Strong problem-solving ability: You can break down unclear problems, form hypotheses, design experiments and make decisions using evidence.
- Engineering judgement: You understand that not every problem requires an LLM, agent or vector database. You choose the simplest reliable approach that meets the requirement.
- Curiosity with depth: You learn new tools quickly, but you do not follow frameworks blindly. You care about understanding how and why systems work.
- Measurement-oriented thinking: You do not describe AI systems only as working well. You define test cases, establish baselines and measure improvements.
- Clear communication: You can explain your reasoning, technical decisions, uncertainties and trade-offs to both technical and non-technical stakeholders.
- Comfort with ambiguity: Many applied-AI problems do not begin with clean requirements or perfect datasets. You are comfortable creating structure where it does not yet exist.
- Bias toward shipping: You balance experimentation with execution. You know when to continue researching and when to make a practical engineering decision.
Torinit's Ethos
Our guiding principles are the cornerstone of our success, shaping a dynamic environment where innovation, quality, and continuous growth are not just valued; they're celebrated. As you consider joining our team, we invite you to explore the values that make us unique:
- Transparent: This isn't about having access to everything; it's about fostering an environment where relevant information is proactively shared and easily accessible to our team and our clients. It is a commitment to ensuring information flows freely, enabling everyone to contribute effectively and make well-informed choices.
- Resourceful: We embrace creative thinking and encourage our team to explore inventive ways to overcome obstacles while considering the most efficient use of available resources. This is the key to our ability to excel in any circumstance.
- United: Our success is recognizing that what we do is a collective effort. It takes collaboration and teamwork within our teams and with our clients. We strive to do right for each other by creating win-win scenarios that benefit our clients, our team, and our company.
- Excellence: Excellence is the driving force behind everything we do. It's our inherent desire to continuously reach new milestones and to be the best in our field. We never settle for mediocrity; instead, we constantly raise the bar and ask ourselves, What can I do better.
Equal Opportunity
We are a proud equal opportunity employer, dedicated to promoting diversity, equity, and inclusion in our hiring practices and workplace culture, ensuring that all qualified applicants receive consideration for employment without regard to race, color, religion, gender, gender identity or expression, sexual orientation, national origin, genetics, disability, age, or veteran status.
