Three years ago, "AI jobs" mostly meant data scientists and ML engineers. Today, the AI job market has branched into a dozen distinct roles — some highly technical, some not — and the fastest-growing are not always the ones being written about in tech press. Here is what is actually appearing in job postings across India and East Africa right now.
MLOps and AI Infrastructure Engineers
Building a model is one thing. Running it reliably in production — handling traffic spikes, monitoring performance drift, managing data pipelines, versioning models, and maintaining the infrastructure that keeps everything working — is a completely different skillset. MLOps engineers are the people who do this, and demand for the role has grown faster than supply for the past three years.
In India, this role is appearing at scale at both large IT services firms (Infosys, Wipro, TCS all have growing MLOps practices) and at product-led startups. In East Africa, early-stage companies using AI features need someone who can deploy and maintain models without a 20-person infrastructure team. The role adapts to the scale of the organisation, but the underlying skills — Python, containerisation, model monitoring, data pipeline management — are consistent.
AI Product Managers
AI products have characteristics that traditional products do not: they behave probabilistically (not deterministically), they degrade over time as the data they were trained on becomes stale, they can fail in ways that are hard to predict, and they raise ethical questions that pure software products typically do not. Managing these products requires a PM who understands these dynamics.
Indian product companies (Freshworks, Razorpay, Zepto) and East African tech firms (Andela, Cellulant, Africa's Talking) are all looking for PMs who can write specifications that ML teams can act on, evaluate model outputs for product-level quality, and navigate the uncertainty inherent in AI features. The PM who has enough AI understanding to have an informed conversation with the ML team without needing a translator is disproportionately valuable.
AI Evaluators and Red Team Specialists
As companies deploy AI in higher-stakes contexts — customer service, healthcare, financial advice, legal — the need to evaluate whether the AI is actually doing what it claims becomes critical. AI evaluation specialists design and run systematic tests of model performance, identify failure modes, and maintain the evidence base for model quality claims.
Red teaming — deliberately trying to make AI systems fail, produce harmful outputs, or be manipulated — is a specialised subfield that is growing fast as regulation increases. Anthropic, Google, and OpenAI all have red team functions. Increasingly, companies deploying these vendors' models need internal evaluation capacity too. This is a role that requires creativity, domain knowledge, and systematic thinking rather than deep ML expertise.
AI Training Data Specialists
The quality of an AI model is constrained by the quality of its training data. Building high-quality training datasets — selecting sources, curating examples, ensuring diversity and representation, writing guidelines for annotation teams — is skilled work that sits between data engineering and domain expertise. It is increasingly well-compensated and remains significantly less crowded than ML engineering.
For Africa specifically: there is genuine scarcity in high-quality training data that represents African contexts, languages, and use cases. Anyone building expertise in this area is entering a market where demand will grow as AI companies increasingly prioritise models that work for non-Western populations.
Build the skills these new roles require
The AI Agents, Data Science, and Data Analysis tracks at Ramayai cover the foundations that make you competitive for these emerging roles — not just the theory, but the hands-on practice.
Explore AI Agents Track →