I lead People and Talent functions at scale, and I build the agentic AI workflows my teams run on. One of them is running further down this page.
Ten years in talent, most of it spent making the tooling keep up.
Most recruiting leaders manage the funnel. I rebuild it.
Over the last decade I've run global technical talent through hypergrowth, slowdowns, restructures, and PE rolls. I owned the workforce planning behind which roles opened, which ones paused, and which ones went away. Doing that well meant getting at data the ATS wouldn't give me, which meant waiting on engineering bandwidth I was never going to be prioritized for. Eventually I stopped waiting and learned to build the tools myself.
Now I do both jobs. I lead the talent strategy, and I architect the workflows underneath it: sourcing pipelines, scorecard screeners, interview coordination. They're what lets a five-person talent team cover the ground of a fifteen-person one. Every one of them stops at a human checkpoint before anything reaches a candidate, because an agent can do the work but it has no business making the call.
The short version of how I actually think about this job.
A CEO asked me for a hiring metric I couldn't get from the ATS. I built the dashboard that night.
Around the same time it was getting obvious that AI would either replace recruiters or make them dramatically better at the job. I had an opinion about which one, and building the tools was the only honest way to find out if I was right.
I've been doing both jobs ever since. Running the function during the day, building for it at night, more or less.
And a slow, thoughtful process still loses the candidate to whoever moved first. You need both at once, which is more or less the whole reason I build tooling.
Submittals and screens go up the moment anyone asks for them. I'd rather be measured on quality-of-hire, and on whether those hires are still here a year later.
" I learned to code because waiting on engineering bandwidth meant my recruiters waited too. Nobody was coming to fix that for us.
" Time-to-fill looks great right up until you check how those hires are doing at six months. I don't trust either number on its own.
Recommendations from peers, leaders, and team members across the last decade. Pulled verbatim from LinkedIn.
AJ's approach to recruitment is both innovative and impactful. He truly understands the profound responsibility recruiters have in aligning the best talent with companies. His ability to design and execute global hiring — and his guidance on advanced hiring tools — is invaluable.
AJ is a rare breed. He is a software developer, engineer, and technologist who applies his experience to the highest talent acquisition level. You can see his impact in how his team responds to his direction — it's probably why his engagement scores far exceed the industry benchmark.
AJ has an innate ability to capture what motivates each of his team members and help them deliver results. He's a natural teacher, and quickly built credibility with the business given his technical background. AJ would be a great people leader for any Technical Recruiting team.
As Nextiva's Senior Manager of Technical Recruiting, AJ successfully built a top notch team of Technical recruiters in US and Mexico. They achieved incredible results due to AJ's hand-on leadership and coaching. AJ built great relationships with the Development and Engineering Leaders, always surpassing their expectations.
Talent Acquisition is traditionally driven by numbers. The way AJ engages his teams, numbers always follow — and he doesn't have to rely on a spreadsheet. He'll drive to improve any TA function, creating a fast, duplicatable, and highly scalable process.
Occasionally you meet someone who touches your life in a way they probably don't realize. AJ was one of those people — extremely wise and well-versed in the technical world, and an amazing leader. He constantly pushed my team to grow both professionally and personally. Any organization would be lucky to have him.
Three agentic recruiting workflows, each with a hard human checkpoint. Run one below and watch every stage, including the point where it stops and waits for a recruiter.
Choose a workflow and run it. Each stage builds up below. The recruiter approves at the checkpoint.
The same three workflows, with their inputs, tool calls, and where each one stops for a human.
Searches the sources I point it at, ranks profiles against the rubric instead of a keyword string, and drafts outreach for each one.
Review the match scores and approve the outreach drafts before anything sends.
Reads the resume, scores it against the rubric with a cited span behind every score, and flags the dimensions it couldn't find evidence for.
Check the cited evidence against the raw resume before the gate opens.
Finds the overlap across a five-person panel in three time zones, then writes each interviewer's briefing packet.
Authorize the proposed slots before any invitation reaches the candidate.
These workflows feed VISION, a live operations dashboard covering pipeline health, time-to-fill, source of hire, and interviewer load. It's the view leadership reads when they decide what opens, what holds, and what closes.
Explore VISION →
Two disciplines I run at the same time. Most teams need both and end up hiring for one.
Capacity models and headcount forecasts tied to revenue milestones rather than to an aspirational org chart.
Fast, honest, respectful pipelines. The kind people say nice things about even after you turn them down.
Cost-per-hire, source efficiency, pass-through rates, quality-of-hire post-onboarding. The numbers leadership actually asks for.
Objective rubrics, balanced panels, anonymized screening. Structure catches more bias than training does.
Multi-stage workflows that source, score, and draft outreach without the recruiter ever leaving their queue.
Vector embeddings and LLM-based matching surface the right candidates Boolean strings missed for years.
Greenhouse, Lever, Slack, Google Workspace. Custom API and webhook glue so data syncs without manual entry.
Every workflow pauses for recruiter review at each decision gate. The agent drafts, the recruiter decides, and I haven't shipped one that works any other way.
The roles that built the thesis. Full PDF and LinkedIn for the complete history.
Leading global recruiting through a 139% scale-up while keeping the team's hiring rubric, infrastructure, and tooling in lockstep with the engineering org.
Ran the same workflow design at three PE-backed companies, largely to find out how much of it was portable between very different orgs. Most of it was.
The role where the argument got settled. I replaced external agency spend with an in-house automated sourcing engine, then ran the numbers in front of the leadership team.
Took weekly regional office revenue from $10K to $57K in 18 months. Generated $800K in business revenue through pricing automation and 120 active contingent placements.
First TA director seat. Built the daily sourcing and feedback systems that ran the team, beat TTF and offer-acceptance benchmarks, and kept every recruiter I hired.
Where I learned to build tools instead of waiting for them, writing JavaScript query builders and scraping utilities for the team. Generated $1.4M in gross profit. President's Club and Boolean Sourcing Competition honors.
What recruiters and hiring leaders most often ask about this work.
Standard automation runs a fixed script. An agentic workflow can reason, branch, retry, and call its own tools, the way a recruiter knows when to broaden a search, when to flag data they don't trust, and when to stop and ask someone.
I build them with a hard rule: the agent does the work, but the recruiter makes the decisions. Every workflow pauses at a human checkpoint before anything reaches a candidate.
In my experience, yes, for the right roles. Semantic search finds people whose experience maps to a role even when their resume never uses the keywords you'd think to search for. Boolean misses those people entirely.
The bigger gain is further down, on outreach. The agent drafts a personalized note for every candidate it surfaces, and that's what actually moves response rates. Ten people with a real hook will out-perform a hundred on a template every time.
Three things. I redact identifiers (name, photo, address, school) before anything gets scored. I require a cited span of evidence behind every rubric score, so there are no opaque outputs to argue with. And the agent never makes the final call on a candidate, it hands a recruiter something to check.
Between NYC Local Law 144 and the EU AI Act, this is also turning into a compliance question, so it's worth building the guardrails in before you need them.
If you need an engineer to own and ship a production system, I'm the wrong hire. I'm fluent enough to direct a build and review what comes back, not to be the person maintaining it at 2am.
I'm the right hire when you need a strategic TA leader at the top of the function: someone who can architect the workflows, set the rubrics, and brief the engineers, while still spending most of their time on the parts of recruiting that only work human to human.
Always happy to talk talent strategy, recruiting operations, and workflow design.
Based in Dallas, TX. Work with teams remotely across US, LATAM, and India time zones.