Nouryon
Website:
nouryon.com
Job details:
AI Solutions Analyst
📍 Mumbai Office
About the job
At Nouryon, our global team of changemakers takes positive action every day, to reach higher goals collectively and individually. We create innovative and sustainable solutions for our customers to answer society's needs – today and in the future.
The AI Solutions Analyst is a developer on Nouryon's AI invoice intelligence platform — an application that reads supplier invoices and extracts charge-level detail to give the business full transparency into logistics cost. This role owns the end-to-end onboarding of vendors onto the platform — analysing each vendor's invoice format, engineering the AI instructions that teach the system to read it correctly, and validating extraction quality until it can be trusted. This role is the connective tissue between the digital transformation team, logistics and procurement stakeholders, and the data that drives cost-leakage decisions — ensuring that AI-extracted invoice data translates into measurable savings, organisational transparency, and continuous improvement.
Job Description
The role is centered around the following core pillars:
- AI Prompt Engineering – Design, tune and test the natural-language instructions that teach the AI to read each vendor's invoice format accurately.
- Vendor Onboarding – Own the end-to-end onboarding of new vendors onto the platform, from first sample invoice to live, validated extraction.
- Data Analytics – Serve as the analytical engine of the team by measuring extraction accuracy, error patterns, and AI processing cost, and translating findings into targeted improvements.
- Data Quality Assurance – Validate AI output against source documents and safeguard the integrity of the invoice data reaching the business.
- Stakeholder Management – Coordinate with the digital transformation team, logistics procurement, operations, and vendors to align formats, priorities and expectations.
- Process Improvements – Identify and deploy improvements in onboarding, extraction, and quality-assurance workflows.
Job Responsibilities
Vendor Onboarding & Format Analysis
- Collect and analyse representative sample invoices from new vendors across warehouse, ocean freight, road/rail, and customs brokerage modes.
- Identify each vendor's invoice structure, charge-line conventions, and formatting quirks that affect automated extraction.
- Onboard vendors onto the platform, configuring extraction profiles and routing settings per vendor.
- Own a named portfolio of vendors end-to-end, from onboarding through to sustained extraction quality.
AI Prompt Engineering & Model Training
- Design, write and iterate custom extraction prompts that teach the AI to read each vendor's invoices correctly.
- Test prompt changes against curated reference sets before releasing them to production.
- Diagnose extraction failures to root cause and resolve them at the prompt level, permanently rather than case by case.
- Continuously improve extraction accuracy and reduce AI processing cost per invoice.
Validation & Data Quality
- Validate AI-extracted charge lines, invoice metadata and classifications against source documents until extraction holds reliably.
- Approve, correct or reject extracted data, ensuring only verified information reaches downstream reporting.
- Curate golden test sets from verified invoices to support regression testing.
- Safeguard data quality, completeness and consistency across the vendor portfolio.
Analytics & Performance Reporting
- Measure and report extraction accuracy, first-pass approval rate, and AI cost per invoice for owned vendors.
- Analyse error patterns across vendors, modes and document types to identify systemic improvement opportunities.
- Translate analytical findings into clear, actionable recommendations for the programme team and senior stakeholders.
- Prepare periodic performance reporting for leadership.
Stakeholder Engagement & Process Improvement
- Act as a key interface between the digital transformation team, logistics procurement, operations, and vendors on onboarding and extraction topics.
- Escalate invoice formats or system limitations that require platform-level changes, with a recommended course of action.
- Identify and implement process improvements across onboarding, prompt engineering, and quality-assurance workflows.
- Contribute to the continuous improvement of analytics, reporting, and decision-support tooling.
We believe you bring (Education & Experience)
This is a specific profile. We are not looking for a generalist analyst — we are looking for someone who has built AI solutions in the field, against real data, with users watching.
1. Forward Deployed Engineer (FDE) experience
You have worked embedded with a business function, not behind a ticket queue. You have taken an ambiguous operational problem, built a working solution against real data, and iterated it in front of the people who use it. You are comfortable when requirements arrive as a conversation rather than a specification.
2. Prompting
You have built something with an LLM that worked measurably better on Friday than it did on Monday — and you can explain exactly why. You treat prompts as engineered artefacts: versioned, tested against a reference set, and improved against a number. Experience with structured extraction from documents is a strong advantage.
3. Scrum
You have delivered in sprints — planning, stand-up, demo, retro — in a self-organising team. This team has no team lead: you will pick up work from a prioritised backlog, commit to it, and be accountable for it without being assigned it.
4. SQL and Python (pandas)
Intermediate SQL: joins, aggregations, grouping, window functions. Python at pandas level, sufficient to analyse extraction results offline, compare model outputs across runs, and quantify accuracy. You use data to prove an improvement, not to describe it.
5. Structured problem decomposition
You turn "this vendor extracts badly" into a specific, testable hypothesis, a defined check, and a fix that holds. You look for the systematic cause behind a single error rather than correcting the symptom.
6. Analytical rigour on documents
You can find one wrong figure in a forty-line charge table — and you are still finding it at four in the afternoon. Sustained precision over high volumes of detailed document work is the daily craft of this role.
Nice to have
- Domain exposure — finance, procurement, logistics or supply chain; invoices, freight, warehousing, customs (Invoice processing, Accounts Payable)
- Data annotation, labelling or data-quality operations background
- Palantir Foundry experience
- Statistical or ML modelling depth beyond descriptive analytics
- BI tooling — Power BI, Tableau
- Scrum certification — CSM / PSM
- Version control — Git fundamentals
- Experience supporting European stakeholders from an offshore or captive centre
Education & Experience
- 2 to 8 years of relevant experience in forward deployed engineering, solution delivery, data analytics, or technical consulting.
- Bachelor's Degree (BA/BS) in Engineering, Computer Science, Data Science, Supply Chain, Operations Research, or a related field from an accredited institution.
Click on Apply to know more.
