Business Integrator and Talent Ops · Berlin

Doris Jovović

I build the systems teams hire, onboard and operate with. And I onboard AI the way I onboard people.

Six years across fast-scaling start-ups, recruitment agencies and EU-funded programmes. I find the root cause, pick the simplest fix that works, and write it down so it keeps working when I am not in the room.

See the work Read the AI Ops Playbook
3 weeks

Time to hire, down from four months, after a new ATS and an in-house process

20+ hrs

Manual talent ops work removed every week

80%

Less spent on headhunters once hiring moved in-house

90%+

Hiring success rate for a talent team I managed, with a 9.1 candidate NPS

60K

Applicants in a talent pool built across four regions

From roles at EIT RawMaterials and Atlas Assistants. Details under Experience.

01 · Selected work

Systems that outlast the project

Each one started with something that only worked because one person held it in their head. The fix was rarely a new tool. It was a clear process, one source of truth, and a check in the right place.

Edition 1 · October 202614 spreads

AI Ops
Playbook

Tort Claim Strategies · BEAT81In daily use

The AI Ops Playbook

How I onboard an AI agent the way I onboard a new hire: a role, an onboarding pack, house rules with reasons, and feedback written down once. Built first at Tort Claim Strategies, rebuilt at BEAT81.

Problem
Every new AI chat started from zero. Context lived in calls, chats and people's heads.
Built
A folder of plain Markdown notes the agent reads first and writes back to, with ranked sources, dated facts and a weekly clean-up.
Result
Corrections become rules instead of repeats, and a new teammate reads the same notes the agent reads.
Claude CodeObsidianMarkdownGranola
Open the playbook
Campaign lifecycleChecks before money moves
  1. Commercial
  2. Intake
  3. Build
  4. Go-live
  5. Delivery
  6. Invoicing
  7. Reporting
Tort Claim Strategies · 2026Live

Operations rebuild for a fully remote team

A US performance marketing company, nine people across four time zones, where one mistake costs four to five figures.

Problem
No runbook, no SOP coverage, and issues caught only after the money had gone out.
Built
The end-to-end lifecycle mapped into one knowledge base, a 25-check early warning framework, and role scorecards tying each routine task to a signed-off SOP.
Result
Checks now run before money moves. Promoted from Executive Assistant to Operations Manager in two months.
SOPsAuditsScorecardsClaude
Pipeline designTwo ATSs → one
30+ stages today
~6 stages, rules carry the nuance
BEAT81 · 2026Designed, rolling out

One recruiting system for a studio a fortnight

Front of House, Shift Lead and Studio Manager hiring for a Berlin fitness scale-up opening around two studios a month.

Problem
Two ATSs, a funnel grown past 30 stages, and numbers computed by hand in spreadsheets.
Built
Requirements and the target process for one pipeline, reporting inside the ATS, GDPR retention by design, and a proactive talent pool that starts sourcing before a vacancy exists.
Result
One candidate record and one set of numbers, so hiring plans rest on data rather than opinion.
AshbyFountainGreenhouseDashboards
Time to hireBefore · after
Before~4 months
After3 weeks
EIT RawMaterials · 2024Shipped

AI-powered ATS for a niche talent market

Hiring for the European minerals and metals sector, where qualified people are scarce and hard to find.

Problem
Four-month hiring cycles, heavy headhunter spend, and hours of manual interview scheduling.
Built
A new AI-powered ATS with automated coordination, in-house screening, and a compliant talent pool of 16,000+ candidates in six months.
Result
Three-week time to hire, 20+ hours of manual work back every week, headhunter costs down over 80%, and a 35% team expansion in a year.
ATS rolloutAutomationSourcingGDPR
Monthly recurring revenueOne year
Start$40K
12 months$220K
Atlas Assistants · 2023 to 2025Shipped

Talent acquisition run as a product

Matching expert Executive Assistants with entrepreneurs, for a US company growing fast.

Problem
Matching quality drove client retention, but hiring ran on manual admin and gut feel.
Built
A custom matching algorithm inside the existing stack, Custom GPTs for the admin, KPIs and OKRs tied to business goals, and a 60,000-applicant talent pool.
Result
Contributed to MRR growing from $40K to $220K in a year with churn under 3%, a 90%+ hiring success rate and a 9.1 candidate NPS.
MatchingCustom GPTsOKRsTeam of 2
Delivery15-person team
PlanVendors, scope
UATTest, fix
Go-liveApproved
HypercareUntil stable
EIT RawMaterials · 2025 to 2026Shipped

A compliant learning platform, from turnaround to go-live

Brought into an eight-figure EU-funded programme as the key operator during a high-risk phase.

Problem
Broken flows, no learner data infrastructure, and strict GDPR, DSA and ZFU requirements.
Built
Learner data, intake forms, trackers and stakeholder dashboards, plus the delivery plan for the LMS across internal teams and external vendors.
Result
Ran UAT, approved go-live and led hypercare until usage and stability targets were hit.
CornerstoneUATComplianceVendors
02 · AI systems

AI with a role, not a prompt

A language model knows nothing about your business unless you put it in front of it, every time. So I give agents what I give a new hire, and a loop that makes every session leave the notes better than it found them.

  1. 01Capture

    Nothing stays only in the chat or the call.

  2. 02Classify

    Stable rule or volatile number? Which note owns it?

  3. 03File and link

    One home per fact. Everything else links to it.

  4. 04Answer

    Rules first, then the topic note, then the live source.

  5. 05Write back

    A correction becomes a written rule, not a repeat.

  6. 06Clean up

    Weekly: merge, re-verify, archive what is replaced.

Two-layer agent manual

TCS · BEAT81

One file describes how I work and travels with me. A second describes each company. Pasted into a new tool or model, it gets the agent working my way in one step.

Calls into knowledge

TCS · BEAT81

Meeting transcripts sync into the notes. Each call becomes one processed note: corrections to the AI summary, decisions, facts and actions, each pushed to where it belongs.

Candidate matching and Custom GPTs

Atlas Assistants

A matching algorithm built from scratch into the existing stack, and Custom GPTs that took the repetitive admin off the talent team.

AI-powered ATS rollout

EIT RawMaterials

Selected, implemented and automated end to end. More than 20 hours of manual talent ops work gone every week.

1
Context before instructions.

Every rule carries its reason, so the agent can judge the case the rule did not foresee.

2
Gaps are answers.

"To be mapped" beats a plausible guess. Volatile facts carry the date they were true.

3
Judgement stays with people.

The agent spots patterns and challenges assumptions, with reasons. A person decides.

The playbook explains the method in fourteen spreads, with interactive examples.

Read the playbook
03 · Experience

Where I have built things

Fitness tech, legal marketing, EU-funded programmes, recruitment agencies and talent marketplaces, in the US and in Europe. Different industries, the same job: make the work run without heroics.

  1. Sep 2026 to nowBerlin

    FOH and Studio Manager Recruiting Manager · BEAT81

    Berlin fitness tech scale-up with Ride and Circuit studios across five German cities.

    • Own high-volume recruiting operations for Front of House, Shift Leads and Studio Managers: funnel design, assessment standards, launch hiring and the handover into onboarding.
    • Redesigning the ATS pipelines and making the case to consolidate all three funnels into one system.
    • Building one recruiting analytics tracker: stage conversion, pass rates, post-probation retention and compliance.
    • Aligning operations, HR, talent acquisition and studio leads on decision rights, scorecards and one assessment standard.
  2. May 2026 to nowRemote · freelance

    Operations Manager · Tort Claim Strategies

    US performance marketing company for mass tort litigation, fully remote across four time zones.

    • Report to the CEO and own the company-wide operations rebuild, from scoping through execution and follow-through.
    • Mapped the campaign lifecycle into one knowledge base, runbook and SOP coverage map where none existed.
    • Designed a 25-check early warning framework and audits that move checks to before the money goes out.
    • Replaced ad hoc hiring, team management and offboarding with structured systems, and enabled the team through onboarding, guides and playbooks.
    Promoted from Executive Assistant in two months
  3. Feb 2025 to Sep 2026Berlin

    Project Manager · EIT RawMaterials

    European consortium advancing the raw materials sector through funding, education and partnerships.

    • Key operator on an eight-figure programme during a turnaround: stabilised operations and built the systems to keep it compliant and scalable.
    • Ops and product lead for the academies: learner data infrastructure, reporting, intake forms, trackers and dashboards.
    • Owned delivery of a GDPR, DSA and ZFU compliant LMS with a 15-person cross-functional team and external vendors.
    • Risk mitigator on a high-profile public grant, surfacing procurement and governance risks early.
  4. Jun 2023 to Oct 2025Austin, US · remote

    Talent Acquisition Manager · Atlas Assistants

    Matches expert Executive Assistants with entrepreneurs.

    • Led the talent acquisition team with two direct reports, building talent acquisition as a product.
    • Contributed to MRR growing from $40K to $220K in one year, with churn below 3%.
    • Raised hiring success above 90% and held a 9.1 candidate NPS.
    • Built a custom matching algorithm, Custom GPTs and a 60,000-applicant talent pool across LATAM, the Philippines, Eastern Europe and South Africa.
    Started as Talent Acquisition Specialist
  5. Apr 2024 to Jan 2025Berlin

    HR Operations · EIT RawMaterials

    • Implemented an AI-powered ATS and automations, removing 20+ hours of manual talent ops work a week.
    • Cut time to hire from four months to three weeks and headhunter spend by over 80%, supporting 35% team growth in a year.
    • Grew a compliant talent pool to 16,000+ candidates in six months in a scarce, niche market.
Earlier: consulting, recruitment operations and research
  1. Jun 2023 to nowRemote

    Talent Acquisition Consultant · Freelance

    • Recruiting operations consulting for US clients, including an Airtable-based ATS built with zero budget: front end, automations and forecasting.
  2. Aug 2021 to May 2023London, UK · remote

    Recruitment Operations Specialist · JobRack

    • Recruitment strategy and sourcing for tech and non-tech roles, KPI tracking, and the SOPs behind recruitment and operations.
  3. Oct 2022 to Apr 2024Potsdam

    Research Assistant · University of Potsdam

    • Experimental linguistics research: experiment materials, documentation and statistical data work for publication.
  4. Jul 2020 to Dec 2021Remote

    International Recruitment · ENGOO

    • Skill-based assessment and mock lessons for a platform with 11,000+ tutors, with honest, developmental feedback for every candidate.
04 · Capabilities

What I bring to a team

R&D and production. The analysis, technical setup and process design behind the scenes, and then sitting with people until it works for them.

01

Business operations

Lifecycle mapping, SOPs and runbooks, audits and early warning checks, vendor delivery, GDPR and grant compliance.

02

Talent operations

High-volume funnels, ATS implementation and administration, interview plans and scorecards, onboarding and offboarding.

03

AI and automation

Agent manuals and knowledge bases, matching logic, Custom GPTs, AI-powered ATS rollouts, no-code automations.

04

Analytics

Funnel and conversion reporting, KPI and OKR design, time to hire and retention analysis, dashboards. Statistics in R and Python.

05

Enablement

Guides and playbooks, handovers, role scorecards, training in 1:1 and group settings.

06

Stakeholder alignment

From recruiters and hiring managers to founders and executives: decision rights, one standard, follow-through.

Recruiting systems

AshbyFountainGreenhouseTeamtailorWorkablePersonioTestGorilla

AI and knowledge

Claude CodeClaudeChatGPTObsidianNotionGranola

Data and ops

AirtableBaserowCornerstoneTypeformGoogle WorkspaceSlackRPython

Education

  • MSc Linguistics, Empirical and Theoretical FoundationsUniversity of Potsdam, Germany
  • Specialist Degree, English Language and LiteratureUniversity of Montenegro, teacher training programme
  • BA English Language and LiteratureUniversity of Montenegro

Languages

  • English
  • Croatian
  • Serbian
  • German
  • Russian
  • Spanish

Let's compare notes.

Building business or talent operations, untangling a process that only works when one person is in the room, or teaching AI how your business runs? I am happy to talk it through.