AI Engineering Job Titles Explained: Which One Fits You

If you search for AI engineering jobs, the titles stop being useful quickly. As of 4 October 2026, AgenticCareers.co tracks 988 distinct AI roles at 315 companies, counting a role posted in several cities once and leaving out non-AI roles at AI companies. 104 of them have "AI Engineer" in the title. When we sort those 104 by the work the posting describes, 34 are agent engineering, 16 are machine learning engineering, 13 are LLM engineering, 6 are solutions roles and 5 are forward deployed roles, with the rest spread across other categories.

So the title tells you little. The six families below are the ones worth telling apart.

The six families, by the numbers

• AI Agent Engineer: 182 roles at 90 companies. Builds systems where a model plans, calls tools and completes multi-step tasks. The largest category on the board, and senior-heavy: 127 of 182 are senior level or above. Median advertised range $177K to $263K (54 listings with disclosed pay).
• AI Solutions Engineer: 117 roles at 38 companies. Customer-facing technical work: scoping, demos, proofs of concept and integration design. More mid-level openings than the other families (54 of 117). Median advertised range $153K to $222K (36 listings).
• AI Infrastructure: 91 roles at 24 companies. Training and inference systems, serving and compute. Concentrated in a few employers and the most senior family: 69 of 91 are senior level or above. Median advertised range $196K to $313K (13 listings).
• ML Engineer: 80 roles at 43 companies. Trains, evaluates and ships models, often outside the LLM world: ranking, fraud, forecasting. Median advertised range $197K to $262K (31 listings).
• LLM Engineer: 57 roles at 47 companies. Builds product features on language models: prompting, retrieval, evaluation, fine-tuning. Spread thinly, about one role per company. Median advertised range $193K to $243K (12 listings).
• Forward Deployed Engineer: 30 roles at 16 companies. An engineer embedded with a customer, writing production code on the customer's systems. The only family here where most openings are mid-level (19 of 30). Median advertised range $184K to $320K (15 listings).

How to tell which job a posting really is

Read the verbs in the responsibilities, not the title.

• Orchestrate, tool use, multi-step, autonomy: agent engineering.
• Train, features, model performance, experiments: ML engineering.
• Prompts, retrieval, evals, fine-tune: LLM engineering.
• Customers, demos, proofs of concept, pre-sales: solutions engineering.
• Embedded with the customer, deploy in their environment, travel: forward deployed.
• GPUs, inference, serving, latency, clusters: infrastructure.

Which one fits your background

• Backend or full-stack engineer: agent engineering and LLM engineering are the closest moves. See what an agentic AI engineer does and what an LLM engineer does.
• Data scientist or ML background: ML engineering, or LLM engineering with a focus on evaluation. Our agent engineer vs ML engineer comparison covers the difference.
• Engineer who likes working with customers: forward deployed or solutions roles. Start with what a forward deployed engineer is.
• Platform, SRE or distributed systems: AI infrastructure.

Two things the numbers say about getting in

Entry-level roles are rare. Only 23 of the 988 roles are junior or intern positions. Most people enter these families by moving sideways from an existing engineering job, not by applying to a junior AI title.

Forward deployed and solutions roles are the most open to mid-level engineers. 19 of 30 forward deployed roles and 54 of 117 solutions roles are mid-level, against 52 of 182 in agent engineering.

How we counted

Categories are assigned from the content of each posting by an automated classifier, so a small share will be wrong. Pay figures use only listings that disclose a yearly USD range; each figure is the median of the low ends and the median of the high ends, and covers advertised salary, not equity or bonus. Live counts for every category are on the roles page, and the monthly figures are published as an open dataset on our data page.